Primary tumour asphericity on baseline 18F-FDG PET/CT provides incremental prognostic value beyond stage and metabolic burden in breast carcinoma: association with metastasis, early metabolic response, and survival outcomes

Primary tumour asphericity on baseline 18F-FDG PET/CT provides incremental prognostic value beyond stage and metabolic burden in breast carcinoma: association with metastasis, early metabolic response, and survival outcomes | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Primary tumour asphericity on baseline 18F-FDG PET/CT provides incremental prognostic value beyond stage and metabolic burden in breast carcinoma: association with metastasis, early metabolic response, and survival outcomes Nitin Gupta, Amit Rana, Manpreet kaur This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8862488/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Tumour sphericity (TS) and its inverse, tumour asphericity (ASP) = 1 - TS, derived from baseline 18 F-FDG PET/CT may capture morphologic heterogeneity linked to tumour aggressiveness. Aim To determine whether baseline primary tumour asphericity on 18 F-FDG PET/CT provides incremental prognostic value beyond clinical stage, molecular subtype, histologic grade, and PET metabolic burden (SUVmax, MTV, TLG) for synchronous distant metastasis and survival outcomes in breast carcinoma. Methods This retrospective study included 202 women (median age 52 years) with biopsy-confirmed breast cancer staged by 18 F-FDG PET/CT (January 2017 to December 2021) and followed for a median of 48 months. Primary endpoints were synchronous distant metastasis at presentation and overall survival (OS); early metabolic non response and progression-free survival (PFS) were secondary endpoints. Primary tumours were semi automatically segmented (Python; Trimesh) and TS/ASP were computed from 3D surface geometry. Optimal cut offs were identified by ROC analysis. Early metabolic non response was defined as failure to achieve ≥ 30% reduction in primary tumour SUVmax on interim PET/CT (where available). Multivariable logistic and Cox models were adjusted for age, TNM stage, molecular subtype, tumour grade, and PET metabolic metrics (SUVmax, MTV, TLG). Results Median TS was 0.78 (IQR 0.72–0.84); an ASP threshold of 0.25 optimally separated metastatic from non-metastatic cases (AUC 0.78, 95% CI 0.71–0.84). High ASP (> 0.25) independently predicted synchronous distant metastasis (OR 3.1, 95% CI 1.6-6.0; p = 0.001) and was associated with inferior 5 year OS after adjustment (HR 2.4, 95% CI 1.2–4.7; p = 0.010). High ASP (> 0.25; TS < 0.75) increased the likelihood of early metabolic non response (OR 2.5, 95% CI 1.4–4.5; p = 0.002). Associations with PFS were also observed (HR 1.9, 95% CI 1.2–2.9; p = 0.004). In decision curve analysis for 5 year PFS, models including ASP showed higher net benefit than clinical only models across clinically relevant thresholds. Conclusions Baseline tumour asphericity is a reproducible, interpretable biomarker of metastatic dissemination and adverse survival in breast cancer, complementing stage, tumour biology, and metabolic burden. Its association with early metabolic non response supports a role for PET/CT based risk stratification and treatment tailoring, pending multicentre validation. Breast cancer 18F-FDG PET/CT radiomics tumour asphericity metastasis survival Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Breast cancer is the most commonly diagnosed malignancy in women worldwide and remains a leading cause of cancer mortality[ 1 ]. Despite major advances in screening, systemic therapy, and multidisciplinary care, outcomes remain heterogeneous, ranging from long term cure in localized disease to early relapse and death in biologically aggressive or disseminated presentations [ 2 ]. Because prognosis and treatment selection depend strongly on stage and tumour biology, accurate baseline risk stratification is central to individualized treatment planning [ 2 ]. However, routine clinicopathologic variables (tumour size, grade, receptor status, nodal involvement) may not fully capture spatial and metabolic complexity that could underlie early dissemination and variable response to therapy. Baseline staging investigations for distant metastasis are generally not recommended for asymptomatic stage I-II disease but are considered for stage III because of the higher prevalence of occult metastatic spread and the potential for management changing findings [ 3 ]. Whole body 18 F-FDG PET/CT provides integrated anatomic and metabolic assessment and is increasingly used for intermediate and high risk breast cancer, where it can upstage disease, guide biopsy of equivocal lesions, and inform systemic and locoregional treatment strategies [ 4 ]. For quantitative assessment, conventional PET biomarkers, SUVmax, metabolic tumour volume (MTV), and total lesion glycolysis (TLG), capture uptake intensity and tumour burden but only partially reflect intra tumoural spatial complexity and morphologic irregularity. Systematic reviews and meta-analyses support prognostic associations for MTV/TLG in breast cancer, but effect sizes vary across studies and meaningful overlap persists among patients with similar clinicopathologic features. Robust quantification also depends on harmonized acquisition and reconstruction, as emphasized by contemporary procedure guidelines [ 5 , 6 ]. Treatment response on PET is commonly summarized using established metabolic response frameworks (e.g., the 1999 EORTC recommendations and PERCIST), which largely reduce complex spatial change to summary measures of uptake or lesion burden [ 7 , 8 ]. Consequently, there is interest in complementary descriptors that capture tumour geometry and heterogeneity and may improve risk stratification at baseline and during therapy. Radiomics extends quantitative imaging by extracting high dimensional descriptors of intensity distribution, texture, and geometry that may act as imaging surrogates of tumour phenotype and microenvironment [ 9 , 10 , 11 ]. Standardization initiatives such as the Image Biomarker Standardisation Initiative (IBSI) provide consensus definitions and reporting guidance that can improve reproducibility across software implementations and institutions [ 12 ]. Among radiomic feature families, shape metrics are attractive because they are intuitive and comparatively less dependent on intensity normalization. Asphericity, defined as 1 minus sphericity, quantifies deviation from an ideal sphere and can be derived from three dimensional lesion geometry. In FDG avid malignancies such as non small cell lung cancer and head and neck cancer, pretherapeutic asphericity has demonstrated independent prognostic value beyond standard metabolic indices [ 13 , 14 ]. In breast cancer, PET derived radiomic and volumetric signatures (including texture features) have shown prognostic relevance, but the specific contribution of baseline primary tumour asphericity has not been systematically evaluated in routine clinical cohorts [ 15 ]. Conceptually, asphericity summarizes how “irregular” a metabolic tumour volume appears in three dimensions, integrating boundary complexity into a single scalar value. Because it is computed from the lesion’s surface area and volume, it can be implemented with modest computational overhead once segmentation is available. This practical profile makes ASP a candidate for inclusion in structured PET/CT reports alongside conventional metrics, provided segmentation and acquisition are standardized. Using a single institutional cohort of women with histologically confirmed breast carcinoma who underwent baseline 18 F-FDG PET/CT prior to systemic therapy, we tested whether primary tumour asphericity is associated with synchronous distant metastasis (primary endpoint) and overall survival, and secondarily with early metabolic non response and progression free survival, and whether it adds incremental prognostic value beyond clinical stage and conventional PET metrics [ 4 ]. Why this matters clinically: In routine practice, staging and treatment intensity in breast cancer are driven mainly by clinical stage and tumour biology, yet patients with similar stage and metabolic burden can diverge markedly in metastatic risk and outcomes. A simple shape descriptor derived from the primary tumour metabolic volume, such as asphericity, can help identify biologically aggressive disease with a higher likelihood of occult dissemination and early non response. If validated prospectively, reporting ASP alongside stage, subtype, and MTV/TLG could support risk adapted management, including triaging borderline stage II-III patients for comprehensive metastatic work up, guiding early escalation or switch of systemic therapy in predicted non responders, and tailoring surveillance and clinical trial enrolment. Materials and Methods Study Population This retrospective study was approved by the institutional review board with waiver of informed consent. Clinical records were screened to identify consecutive patients with newly diagnosed, biopsy confirmed breast carcinoma who underwent baseline 18 F-FDG PET/CT between January 2017 and December 2021 and received standard first line treatment (neoadjuvant chemotherapy [NAC] with or without targeted therapy, or upfront surgery). For patients receiving NAC, interim PET/CT was typically performed after two cycles of chemotherapy for early metabolic non response assessment; exact imaging intervals were not uniformly recorded. All patients had clinical follow up for survival endpoints. Exclusion criteria: Multifocal primary tumours, prior malignancy, incomplete imaging, or follow up < 6 months. PET/CT Acquisition All studies were acquired on a GE Discovery IQ 610 (32 slice) hybrid PET/CT system (GE Healthcare, Waukesha, WI, USA). Patients received ~ 3.7 MBq/kg of 18 F-FDG, and imaging was performed after an uptake time of 60 ± 10 minutes. Whole body PET emission data were acquired in 3D mode using a multi bed position protocol (2 minutes per bed position) with bed overlap as per institutional protocol. PET images used for analysis were reconstructed using Q.Clear (BSREM; DICOM reconstruction method QCHD) with CT based attenuation correction, on a 192 × 192 matrix with voxel size 3.65 × 3.65 × 3.23 mm (reconstruction diameter 700 mm). Low dose CT for attenuation correction and anatomical localization was performed at 120 kVp with automatic tube current modulation, reconstructed at 512 × 512 with 0.85 × 0.85 mm in plane pixel spacing and 1.25 mm slice thickness (pitch 0.9375, rotation time 0.7 s; reconstruction diameter 434 mm). Multiplanar reformats (axial, coronal, sagittal) were generated on the workstation. Image Processing Tumour segmentation, surface area, volume, and asphericity calculations Tumour segmentation and three dimensional (3D) geometric analysis were performed in Python (v3.10) using NiBabel, scikit-image, SciPy, and Trimesh. Whole body PET/CT studies were converted from DICOM to NIfTI (PET_converted.nii.gz) using dicom2nifti, preserving image orientation and voxel geometry. To reduce off-target seed placement and constrain the segmentation search space, segmentation was spatially constrained using a CT derived thoracic body mask (HU −500 to +1000; morphological closing radius 20 voxels) encompassing the thoracic field (including the breasts) while excluding abdominal and pelvic physiologic uptake. TotalSegmentator (v2.12.0) [16] and nnU-Net (v1.7.1) [17] were used to generate organ and axillary lymph node (ALN) exclusion masks to support accurate primary breast tumour delineation. Lesions were delineated using the scikit-image random walker algorithm with automated seed initialization. Foreground seeds were sampled from high-uptake voxels within the constrained PET volume using a foreground candidate threshold of 5% of SUVmax, while background seeds were sampled from low-uptake voxels using a 2% of SUVmax threshold (beta = 5; solver tolerance = 1×10⁻²). These thresholds were used only to define candidate seed pools for robust initialization and were not applied as intensity cut offs for the final tumour boundary. The final tumour label was exported as a binary mask and underwent post processing (morphological closing radius 2 voxels, hole filling for cavities ≥500 voxels, and boundary regularization) to ensure continuity and topological consistency. All post processing was applied only to the binary label mask and did not modify PET voxel intensities. Surface mesh extraction For surface based shape analysis, the tumour mask was resampled to an isotropic 1 mm grid using nearest neighbour interpolation (label preserving) and a triangular surface mesh was extracted using the Marching cubes algorithm (skimage.measure.marching_cubes). The resulting vertices and faces were imported into Trimesh to compute enclosed tumour volume (V, mm³) and tumour surface area (A, mm²). Tumour sphericity (Ψ) and asphericity (ASP) were calculated as: Ψ = π (1/3) (6V) (2/3) / A, and ASP = 1 - Ψ, where V is tumour volume (mm³) and A is tumour surface area (mm²). Quality control of tumour segmentation QC overlays were generated for every case by superimposing the binary contour on co-registered PET uptake in three orthogonal planes at the lesion centroid (Mask_QC.py; matplotlib, scikit-image) and saved as high resolution images. Two nuclear medicine physicians independently reviewed all overlays. Segmentations were flagged if the contour failed to enclose metabolically active tumour, included adjacent non tumour uptake, or missed obvious extensions/satellite foci. A CT based thoracic body mask (HU -500 to +1000; morphological closing radius = 20 voxels) was applied before seeding to restrict seed placement to the thoracic field and avoid abdominal/pelvic physiologic uptake. When QC suggested misplaced or insufficient seeding, a manual seed coordinate (seed_xyz) was placed within the primary breast lesion and segmentation was repeated (<5% of cases). Mask voxel count and volume were logged; outliers were re-reviewed. Only masks passing visual and quantitative checks were included for mesh extraction and VOI metrics, and all QC overlays were archived. Reproducibility was assessed by repeat segmentation in a subset, and ASP agreement was summarized using the intraclass correlation coefficient (ICC). Interobserver ICC (intraclass correlation coefficient) for ASP was 0.94 (95% CI 0.90-0.97). Image 1 (from patient # 1) show example of a left primary breast tumour segmetation from whole body FDG PET/CT scan DICOM images. Image 2 (from patient # 2) shows segmentation of a right primary breast tumour from whole body FDG PET/CT scan DICOM images in another patient. Image 3 shows surface mesh 3D rendered orthogonal images of the same tumors from patient 1 and patient 2 respectively processed and generated with meshlab software (MeshLab 2025.07). VOI Statistics (SUV, MTV, TLG) To derive SUV based VOI metrics, raw PET values were converted to SUV units using DICOM metadata (Radiopharmaceutical Information Sequence) and patient weight: SUV(x,y,z) = (PETraw × RescaleSlope + RescaleIntercept) × Wg / DBq, where Wg is patient weight in grams and DBq is injected dose in Bq per scanner metadata. Importantly, VOI metabolic parameters were computed from the native resolution SUV PET image to preserve quantitative accuracy; PET intensities were not smoothed, filtered, or resampled for SUV calculations. To avoid misalignment effects, VOI masks used for quantification were maintained on the native (coarse) PET grid; where mapping between grids was required, masks were transferred using nearest neighbour resampling. PET derived quantitative metrics included SUVmax (maximum SUV voxel within the tumour VOI), SUVpeak (mean SUV within a fixed 1 cm³ spherical volume positioned to maximize local mean uptake within the tumour VOI), and SUVmean (mean SUV of all voxels within the tumour VOI). MTV (cm³) was calculated as the number of tumour voxels multiplied by native PET voxel volume (mm³) divided by 1000, and TLG (g·cm³) was calculated as SUVmean × MTV. All steps (conversion, segmentation, label processing, resampling for mesh, mesh metrics, and VOI metrics) were executed in a fully automated pipeline using fixed parameters applied uniformly across cases. To verify that mesh reconstruction did not influence SUV quantification, SUV metrics were derived exclusively from the native SUV image, whereas mesh derived measures were computed only from the label mask. Endpoints Study endpoints comprised synchronous metastatic disease at presentation, early metabolic non response on interim PET/CT, overall survival (OS), and progression free survival (PFS). Synchronous (de novo) distant metastasis was defined as distant metastatic disease on baseline PET/CT, supported by correlative imaging and/or histopathology when available (histopathologic confirmation in n = 57); equivocal findings were adjudicated using follow up imaging and clinical documentation. Early metabolic response on interim PET/CT (performed for neoadjuvant chemotherapy patients where available, typically after two cycles) was defined a priori as a 30% reduction in primary tumour SUVmax relative to baseline, analogous to the PERCIST 1.0 partial metabolic response threshold but applied to SUVmax in this retrospective cohort. Early metabolic non response was coded as failure to achieve this ≥ 30% SUVmax reduction. %DeltaSUVmax = [(SUVmax_baseline - SUVmax_interim) / SUVmax_baseline] x 100. Overall survival (OS) was defined as the time from baseline PET/CT to death from any cause; patients were censored at last known follow up. PFS was defined as the time from baseline PET/CT to the earliest of (i) first documented disease progression or (ii) death from any cause, whichever occurred first; patients without an event were censored at last event free follow up. Follow-up imaging was not protocol mandated and was typically performed as PET/CT at clinical suspicion of recurrence. Accordingly, progression was determined primarily by the first PET/CT report describing unequivocal progressive disease (e.g., new FDG avid lesions consistent with metastasis and/or clear interval increase in extent or intensity of previously documented malignant uptake), with correlative imaging and/or histopathology used when available. For ROC analyses, OS and PFS were evaluated as 5 year endpoints (event within 5 years vs event free/censored beyond 5 years). Statistical Analysis Optimal ASP cut off for metastasis detection was derived by ROC analysis (Youden index). The primary endpoint was synchronous distant metastasis at presentation (logistic regression); OS was the key time to event endpoint (Cox proportional hazards), with early metabolic non response and PFS analysed as secondary endpoints. Multivariable models adjusted for age, TNM stage, molecular subtype, tumour grade, and PET metabolic metrics (SUVmax, MTV, TLG). Collinearity among continuous predictors was assessed using variance inflation factors; ridge-penalized Cox modelling (L2) was used as a sensitivity analysis to obtain stable estimates when evaluating overlapping metabolic indices. Kaplan-Meier curves compared OS and PFS by ASP strata (log-rank test), and proportional hazards assumptions were checked using Schoenfeld residuals. Decision curve analysis for 5 year PFS compared net benefit of clinical-only, ASP only, metabolic only, and combined models across threshold probabilities 0 - 0.50 versus treat none and treat all strategies (apparent performance; no optimism correction). Statistical analyses were performed in MiniTab version 22.1(Minitab, LLC., PA, USA) and MedCalc (version 23.0.9, MedCalc Software Ltd, Ostend, Belgium) ; ridge sensitivity analyses were performed in Python. P value <0.05 was considered statistically significant. Results Patient Characteristics Of 245 screened patients, 202 met inclusion criteria. Median age was 52 years (range 28-81); 61% were hormone receptor positive/HER2-negative, 23% HER2-positive, and 16% triple negative. Median primary tumour SUVmax was 12.4 (IQR 8.7-17.9), and median ASP was 0.22. Interobserver intraclass correlation coefficient (ICC) for ASP was 0.94 (95% CI 0.90-0.97). Demographic, clinical, and metabolic metrics are shown in Table 1. Primary analyses focused on synchronous distant metastasis at presentation and overall survival; early metabolic non response and progression free survival were evaluated as secondary endpoints. Association between TS, ASP and Baseline Metastasis ROC analysis yielded an optimal TS threshold of 0.75 (equivalent ASP 0.25; AUC 0.78, 95% CI 0.71-0.84). Metastatic disease was present in 52% of high ASP (>0.25) versus 22% of low ASP tumours (p<0.001). In multivariable logistic regression, high ASP independently predicted metastasis (OR 3.1, p = 0.001) along with tumour size (OR 1.8 per cm, p = 0.02). Tumour ASP and Metabolic Treatment Response Interim PET/CT was available for 157 patients receiving NAC. Early metabolic non response was observed in 32% of low ASP tumours versus 59% of high ASP tumours (p = 0.002). Baseline ASP inversely correlated with percentage SUVmax reduction (r = -0.50, p 0.25; corresponding five year PFS was 64 % vs 42 % respectively. High ASP remained an independent predictor of OS (HR = 2.4, p = 0.01); an association with PFS was also observed (HR = 1.9, p = 0.004) after adjustment for age, TNM stage, molecular subtype, tumour grade, SUVmax, MTV and TLG. ROC Analysis The ROC panel (Figure 1) based on tumour asphericity (ASP; ASP = 1 - TS) demonstrated good discrimination across all pre-specified endpoints, with AUCs ranging from 0.78 to 0.81. Discrimination was strongest for 5 year progression free survival (AUC = 0.81) and early metabolic non response (failure to achieve ≥30% SUVmax reduction) (AUC = 0.80), followed by 5 year overall survival (AUC = 0.79) and synchronous baseline metastasis (AUC = 0.78). At a false positive rate of 0.20, sensitivities remained ≥ 0.60 across endpoints, indicating that most events could be identified while limiting false positive classifications. Comparison of tumour asphericity with clinical Stage, histopathology, grade and metabolic Burden Comprehensive multivariate models were constructed to benchmark tumour asphericity (ASP) against conventional prognosticators. Metastatic disease at presentation In an 8 variable logistic regression high ASP (> 0.25, corresponding to TS <0.75) retained the strongest independent association with distant metastasis (adjusted OR = 3.1, 95% CI 1.6-6.0, p = 0.001), exceeding the effect sizes of advanced clinical stage (OR = 2.2 for stage III; OR = 5.8 for stage IV) and metabolic volume (OR = 1.12 per 10 mL MTV). Early metabolic non response In multivariable logistic regression, baseline high asphericity (ASP > 0.25; TS < 0.75) was associated with a 2.5 fold higher odds of early metabolic non response (p = 0.002), independent of stage, subtype, grade, SUVmax, MTV and TLG. Overall and progression free survival In Cox models, ASP remained a significant predictor of OS (HR = 2.4, p = 0.01); an association with PFS was also observed (HR = 1.9, p = 0.004) after simultaneous adjustment for all covariates, ranking second only to stage IV disease. Tumour asphericity (ASP) and its relationship to baseline metastatic status, metabolic treatment response, and change in FDG uptake are shown as box plots and scatter plots in Figure 3. Kaplan-Meier survival curves stratified by tumour asphericity (ASP > 0.25 vs ≤ 0.25) for breast carcinoma in general and for different subtypes are shown in Figure 4 and Figure 5, indicating statistically significant differences in survival between subgroups. These comparisons underscore that morphometric heterogeneity captured by ASP offers additive prognostic information beyond tumour burden and molecular phenotype. Decision curve analysis for 5 year PFS (Figure 6) showed greater net benefit for ASP only and metabolic only models compared with the clinical only model, while the combined model (clinical + ASP + metabolic metrics) provided the highest net benefit across most clinically relevant thresholds (0 - 0.50) relative to treat none and treat all strategies. Independent determinants across endpoints Multivariable models confirmed that morphometric, staging, histologic, and metabolic variables each contributed independent prognostic information. Figure 2 shows Forest plots illustrating the independent effect of high tumour asphericity (ASP > 0.25) alongside clinical and metabolic covariates across four multivariate models. Metastasis: High tumour ASP (> 0.25) and stage IV disease showed the largest independent effects; triple negative phenotype and higher grade were also associated with metastatic presentation. Early metabolic non response: Baseline high asphericity (ASP > 0.25; TS 0.25), stage IV disease and triple negative subtype were associated with inferior OS, with additional contributions from baseline metabolic burden. Progression free survival (secondary endpoint): Similar covariates predicted earlier progression or death, with stage IV disease and ASP showing the largest effect sizes. Discussion In this retrospective cohort, baseline 18 F-PET/CT derived primary tumour asphericity (ASP) was independently associated with synchronous distant metastasis and inferior overall survival, even after adjustment for established clinicopathologic variables and conventional metabolic biomarkers (SUVmax, MTV, TLG). Associations with early metabolic non response and progression free survival were also observed, although PFS ascertainment was clinically driven. Overall, these results support ASP as a simple, interpretable morphometric biomarker that captures complementary information beyond uptake intensity and volumetric burden. A key implication is that tumour geometry on baseline 18 F-FDG PET/CT may capture aspects of aggressiveness that are not fully represented by intensity or volume metrics alone. MTV and TLG reflect tumour burden and have been linked to prognosis in breast cancer, but they do not explicitly quantify the spatial organization of uptake within the tumour volume [ 6 ]. ASP, by contrast, operationalizes geometric irregularity and may summarize complex metabolic contours associated with infiltrative growth, stromal reaction, or heterogeneous viable tumour distribution. Our findings align with prior work in other cancers. In primary non small cell lung cancer and head and neck squamous cell carcinoma, asphericity of pretherapeutic FDG uptake stratified survival independently of tumour volume and uptake intensity [ 13 , 14 ]. Together with these reports, our results support the broader concept that shape based descriptors can provide clinically relevant information across tumour types, even when derived from routine clinical imaging protocols. Breast cancer PET radiomics literature has predominantly emphasized texture and intensity based features, which can reflect metabolic heterogeneity and have been associated with recurrence risk or survival in selected cohorts [ 11 , 15 ]. By focusing on a geometric descriptor that is computationally straightforward, our work complements existing approaches and may be easier to communicate clinically. From an interpretability standpoint, ASP can be described as a single measure of how far the metabolic tumour departs from a sphere, which may support clinician acceptance compared with high dimensional texture signatures. The biological interpretation of ASP is necessarily indirect. A more irregular uptake geometry could reflect a mixture of viable tumour, necrosis, stromal reaction, and heterogeneous perfusion, with uptake contours shaped by local microenvironmental constraints and infiltrative tumour fronts. Hypoxia is a recognized driver of invasion and metastatic dissemination in breast cancer, and hypoxia regulated programs can remodel extracellular matrix and promote tumour cell survival during dissemination [ 18 ]. These mechanisms align with broader frameworks describing how cancers acquire invasive and metastatic capabilities [ 19 ]. Radiomics studies across tumour types further suggest that macroscopic imaging phenotypes can encode aspects of underlying biology and clinical behaviour, providing a rationale for why even simple geometry based features may remain informative after adjustment for intensity and volume metrics [ 10 ]. From a clinical decision standpoint, ASP may help refine which patients are most likely to benefit from comprehensive staging and intensified surveillance. Practice guidance discourages routine baseline metastatic imaging for asymptomatic early stage disease but supports consideration for stage III presentations; within such a framework, ASP could further triage intermediate and high risk patients for targeted work up, confirmatory imaging, or closer follow up when conventional risk indicators yield borderline or discordant signals [ 3 , 4 ]. ASP may also complement established PET response frameworks. While EORTC and PERCIST emphasize changes in uptake intensity (and, in practice, lesion burden), geometry based descriptors may capture a different dimension of response, such as irregular residual uptake or fragmentation, that could be relevant when uptake changes are modest, heterogeneous, or confounded by inflammatory uptake [ 7 , 8 ]. Prospective studies could explore combined response models integrating uptake, volumetrics, and geometry over time, and evaluate whether ASP improves early identification of non responders or patients at risk of early progression. Technical and implementation considerations merit attention. ASP is derived from surface area and volume of a segmented lesion; therefore, it can be sensitive to segmentation strategy, voxel size, smoothing, and partial volume effects, particularly in small lesions or lesions with low contrast. However, because ASP is a shape based metric derived from surface topology, it is theoretically less susceptible to inter scanner SUV calibration errors than intensity based metrics like SUVmax or TLG, provided the segmentation boundary remains topologically consistent. Harmonized acquisition and reconstruction, standardized SUV calibration, and transparent segmentation protocols are essential to reduce measurement variability in PET metrics and derived radiomic features [ 5 ]. Standardized feature definitions and reporting guidance, as promoted by IBSI, further support reproducibility across software implementations and centres [ 12 ]. In practice, adopting a limited set of well defined, clinically interpretable features (e.g., MTV, TLG, ASP) may provide a pragmatic bridge between conventional reporting and more complex radiomic signatures [ 11 ]. From a modelling perspective, improvements in discrimination and net benefit when ASP was added to multivariable models suggest potential utility across clinically plausible decision thresholds. Decision curve analysis is particularly valuable for assessing whether a model improves decision making by balancing harms of false positive actions against missed high risk patients [ 18 , 20 ]. In this study, decision curve analysis was performed for 5 year PFS (a secondary endpoint) to illustrate potential downstream impact on escalation of staging, follow up, or therapy. Future work should focus on multicentre validation with prespecified workflows that include automated or semi-automated segmentation, harmonized acquisition and reconstruction, and prospective evaluation of whether ASP improves clinical decision making when embedded in structured reports. Such studies should also quantify test-retest repeatability and assess sensitivity to segmentation strategy and voxel geometry so that practical reporting thresholds can be defined for routine use [ 5 , 12 ]. Limitations of the study Limitations of the study include its retrospective nature, single centre design introduces potential selection bias and limits generalizability. ASP may vary with delineation strategy, lesion size, and imaging protocol; although we used consistent acquisition and segmentation, external validation should assess robustness across scanners, reconstruction settings, and segmentation paradigms [ 5 , 12 ]. In addition, the study was not designed to link ASP directly to histopathologic correlates such as tumour architecture, necrosis, or hypoxia markers, and such correlative studies would strengthen mechanistic interpretation. Finally, longer follow up and larger cohorts are needed particularly for hormone receptor positive disease where late recurrences are common and treatment decisions are nuanced [ 2 ]. Conclusions Tumour asphericity extracted from baseline 18 F-FDG PET/CT offers an easily obtainable, reproducible biomarker that stratifies risk beyond conventional factors. In our cohort, high asphericity (> 0.25) independently predicted synchronous metastasis and inferior OS after controlling for stage, molecular subtype, grade and metabolic tumour burden; associations with early metabolic non response and PFS were supportive secondary findings. Integration of this simple shape metric into routine PET/CT reports could refine personalised decision making, identifying patients who may benefit from intensified staging, systemic therapy, closer surveillance or inclusion in clinical trials. Prospective multicentre validation and evaluation of dynamic changes during therapy are warranted. Declarations Ethics approval and consent to participate : This retrospective study was approved by the institutional review board with a waiver of informed consent. Consent for publication: Waived due to the retrospective nature of the study. Funding : None. Competing interests: The authors declare no competing interests. Use of AI assisted technologies An AI language model was used to assist with language editing and clarity of the manuscript. All authors reviewed and edited the content and take full responsibility for the accuracy and integrity of the work. No AI tool was used to generate, alter, or analyze patient data, statistical results, or clinical images. Code availability Custom Python code used for tumour segmentation, mesh extraction, and asphericity computation will be made available in a public repository (e.g., GitHub/Zenodo) at the time of publication; during peer review it can be shared with editors and reviewers on reasonable request. Ethics statement: This retrospective study was approved by the institutional review board with a waiver of informed consent. Patient consent for publication: Waived due to the retrospective nature of the study. Funding: None. Conflicts of interest: None declared. Author contributions declarations : Dr. Nitin Gupta (M.B;B.S. M.D. nuclear medicine , Medical Image Analysis (IIT Madras (NPTEL), Certificate in Advanced AI in imaging (RSNA, USA): Conceptualization; Methodology; Software (Python based tumour segmentation and asphericity pipeline); Formal analysis; statistics, visualization; writin original draft; review & editing; Critical revision of the manuscript. Dr. Amit Rana, M.D. ; D.M Medical Oncology: Literature review; manuscript review, editing; Manuscript formatting. Ms. Manpreet Kaur, M.Sc. R.S.O and medical physicist (Nuclear medicine) : Data curation (retrieving patient DICOM data and images and clinical data); Patient follow up and data collection. All authors read and approved the final manuscript and agree to be accountable for all aspects of the work. Ethics approval and consent to participate: This retrospective study was approved by the Institutional Ethics Committee (IEC), Dr. Rajendra Prasad Government Medical College, Kangra (Tanda), Himachal Pradesh, India, with a waiver of informed consent due to the retrospective design and use of anonymized data. References Sung H, Ferlay J, Siegel RL, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209-249. doi:10.3322/caac.21660. PMID:33538338. Waks AG, Winer EP. Breast cancer treatment: a review. N Engl J Med. 2019;381(3):288-300. doi:10.1056/NEJMra1807166. PMID:31483975. Arnaout A, Varela NP, Allarakhia M, et al. Baseline staging imaging for distant metastasis in women with stages I, II, and III breast cancer. Curr Oncol. 2020;27(2):e123-e145. doi:10.3747/co.27.6147. PMID:32489262. Groheux D, Hindie E, Delord M, et al. 18F-FDG PET/CT for staging and restaging of breast cancer. J Nucl Med. 2016;57(Suppl 1):17S-26S. doi:10.2967/jnumed.115.157859. PMID:26834096. Boellaard R, Delgado-Bolton R, Oyen WJG, et al. FDG PET/CT: EANM procedure guidelines for tumour imaging: version 2.0. Eur J Nucl Med Mol Imaging. 2015;42(2):328-354. doi:10.1007/s00259-014-2961-x. PMID:25452219. Wen W, Xuan D, Hu Y, et al. Prognostic value of maximum standard uptake value, metabolic tumor volume, and total lesion glycolysis of positron emission tomography/computed tomography in patients with breast cancer: A systematic review and meta-analysis. PLoS One. 2019;14(12):e0225959. doi:10.1371/journal.pone.0225959. PMID:31826010. Young H, Baum R, Cremerius U, et al. Measurement of clinical and subclinical tumour response using [18F]-fluorodeoxyglucose and positron emission tomography: review and 1999 EORTC recommendations. Eur J Cancer. 1999;35(13):1773-1782. doi:10.1016/S0959-8049(99)00229-4. PMID:10673991. Wahl RL, Jacene H, Kasamon Y, Lodge MA. From RECIST to PERCIST: Evolving considerations for PET response criteria in solid tumors. J Nucl Med. 2009;50(Suppl 1):122S-150S. doi:10.2967/jnumed.108.057307. PMID:19403881. Lambin P, Leijenaar RTH, Deist TM, et al. Radiomics: the bridge between medical imaging and personalised medicine. Nat Rev Clin Oncol. 2017;14(12):749-762. doi:10.1038/nrclinonc.2017.141. PMID:28975929. Aerts HJWL, Velazquez ER, Leijenaar RTH, et al. Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat Commun. 2014;5:4006. doi:10.1038/ncomms5006. PMID:24892406. Ha S, Choi H, Paeng JC, Cheon GJ. Radiomics in oncological PET/CT: a methodological overview. Nucl Med Mol Imaging. 2019;53(1):14-29. doi:10.1007/s13139-019-00571-4. PMID:30828395. Zwanenburg A, Vallières M, Abdalah MA, et al. The Image Biomarker Standardisation Initiative: standardized quantitative radiomics for high-throughput image-based phenotyping. Radiology. 2020;295(2):328-338. doi:10.1148/radiol.2020191145. PMID:32154773. Apostolova I, Rogasch J, Steffen IG, et al. Asphericity of pretherapeutic tumor FDG uptake in primary non-small cell lung cancer: a new prognostic marker. BMC Cancer. 2014;14:896. doi:10.1186/1471-2407-14-896. PMID:25444154. Apostolova I, Steffen IG, Wedel F, et al. Asphericity of pretherapeutic tumor FDG uptake provides independent prognostic value in head-and-neck cancer. Eur Radiol. 2014;24(9):2077-2087. doi:10.1007/s00330-014-3269-8. PMID:24965509. Groheux D, Martineau A, Teixeira L, et al. 18F-FDG PET/CT for predicting the outcome in ER+/HER2- breast cancer patients: comparison of clinicopathological parameters and PET image-derived indices including tumor texture analysis. Breast Cancer Res. 2017;19:3. doi:10.1186/s13058-016-0793-2. PMID:28057031. Wasserthal J, Breit HC, Meyer MT, Pradella M, Hinck D, Sauter AW, et al. TotalSegmentator: Robust segmentation of 104 anatomic structures in CT images. Radiol Artif Intell. 2023;5(5):e230024. doi:10.1148/ryai.230024. PMID:37795137. Isensee F, Jaeger PF, Kohl SAA, Petersen J, Maier-Hein KH. nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nature Methods. 2021;18(2):203-211. doi:10.1038/s41592-020-01008-z. PMID:33288961. 18. Semenza GL. Molecular mechanisms mediating metastasis of hypoxic breast cancer cells. Trends Mol Med. 2012;18(9):534-543. doi:10.1016/j.molmed.2012.08.001. PMID:22921864. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144(5):646-674. doi:10.1016/j.cell.2011.02.013. PMID:21376230. . Vickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models. Med Decis Making. 2006;26(6):565-574. doi:10.1177/0272989X06295361. PMID:17099194. Table Table 1. Showing demographics, imaging & survival metrics of patients included in the study. Section Category n % Median SUVmax (IQR) Median MTV, mL (IQR) Median TLG, g (IQR) Median ASP (IQR) Median PFS (months) Median OS (months) Age group Age 20-29 4 2.0% 9.4 (7.2-12.6) 14.6 (8.2-20.8) 129.6 (70.4-180.5) 0.29 (0.23-0.30) 38 53 Age group Age 30-39 26 13% 11.2(9.8-12.7) 15.8 (9.4-22.7) 136.8 (80.3-190.7) 0.26 (0.19-0.28) 45 60 Age group Age 40-49 56 27.7% 10.6 (8.6-15.6) 15.2 (10.3-23.6) 147.2 (90.6-200.3) 0.24 (0.19-0.29) 46 60 Age group Age 50-59 70 34.7% 12.8 (9.6-16.4) 17.3 (11.6-25.8) 168.2 (100.4-224.7) 0.25 (0.20-0.30) 44 58 Age group Age 60-69 36 17.8% 11.7 (8.6-15.9) 19.3 (12.4-28.5) 173.6 (116.2-232.4) 0.26 (0.21-0.31) 38 52 Age group Age ≥70 10 5.0% 13.6 (11.2-17.4) 20.8 (13.6-30.2) 180.6(120.9-240.5) 0.28 (0.25-0.32) 32 45 TNM stage Stage I 10 5.0% 8.4 (5.7-11.3) 10.7(5.4-15.2) 86.7 (40.7-121.6) 0.18 (0.12-0.22) 60 70 TNM stage Stage II 85 42.1% 10.6 (7.2-14.8) 15.3 (8.7-24.3) 123.6 (62.8-206.2) 0.22 (0.17-0.27) 48 60 TNM stage Stage III 65 32.2% 13.8(10.5-18.3) 22.8 (12.7-34.2) 180.7 (100.4-280.6) 0.30 (0.26-0.35) 36 50 TNM stage Stage IV 42 20.8% 17.6 (14.2-21.7) 28.1 (15.6-38.4) 250.8 (150.4-350.3) 0.33 (0.29-0.38) 20 35 Histopath type Invasive ductal carcinoma (NST) 162 80.2% 12.8 (8.7-16.9) 17.9 (9.6-27.8) 165.7 (82.3-250.6) 0.26 (0.19-0.29) 42 55 Histopath type Invasive lobular carcinoma 12 5.9% 7.2 (5.4-9.8) 14.2 (8.5-22.6) 108.3(50.7-160.4) 0.24 (0.15-0.25) 45 60 Histopath type Mucinous carcinoma 5 2.5% 6.4 (4.1-7.6) 10.2 (5.7-18.6) 80.7 (40.6-144.6) 0.21 (0.13-0.24) 50 65 Histopath type Medullary carcinoma 8 4.0% 11.6 (10.3-18.7 20.4 (12.7-30.8) 184.2 (90.8-282.3) 0.28 (0.23-0.33) 38 48 Histopath type Metaplastic / other special types 4 2.0% 15.3 (12.1-21.8) 24.3 (14.6-35.2) 210.8 (110.6-328.5) 0.30 (0.25-0.35) 30 40 Histopath type Others (papillary, tubular, etc.) 11 5.4% 10.3 (7.2-14.6) 15.2 (8.3-24.9) 120.7 (60.8-204.6) 0.22 (0.17-0.27) 46 58 Tumour grade Grade I 30 14.9% 7.8 (5.3-11.9) 12.4 (6.2-20.5) 84.9 (44.2-137.8) 0.18 (0.12-0.22) 50 65 Tumour grade Grade II 90 44.6% 11.4(8.7-16.4) 19.6 (11.2-30.8) 163.8 (80.6-250.4) 0.22 (0.17-0.27) 44 56 Tumour grade Grade III 82 40.6% 15.6 (11.8-21.4) 23.5 (15.8-37.2) 214.3 (116.5-327.8) 0.28 (0.23-0.33) 34 48 Images Images 1 to 3 are available in the Supplementary Files section. Supplementary Files Image1.docx Image2.docx Image3.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8862488","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":590829986,"identity":"ac67d56c-b054-4ed3-a411-4ace239338d7","order_by":0,"name":"Nitin Gupta","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA0klEQVRIiWNgGAWjYDACZiBOYGCQYWBvALIMLIjRwszYANTCw8BzAKRFgihrGEHG8zBIJIB4RGgxZ+c//uDhjjs85pLPr274USDBwN/enYBXi2Uz0JbEM894LGfnlN3sATpM4szZDXi1GBwGaWk7zGNwOyftBg9Qi4FELrFabp5Ju/mHNC032I/dJtYWwxmJbUC/9OSw3ZYxkOAh7JfzBx98/Nl2R86c/fizm2/+2Mjxt/fi1wIFBxgMGHgMQCweYpTDtLA/IFb1KBgFo2AUjDAAAIs0SHOMrH9dAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0002-9443-9609","institution":"Dr Rajendra Prasad government medical Kangra","correspondingAuthor":true,"prefix":"","firstName":"Nitin","middleName":"","lastName":"Gupta","suffix":""},{"id":590829987,"identity":"1e7973d6-96da-4acd-a440-bf75b0ff476f","order_by":1,"name":"Amit Rana","email":"","orcid":"","institution":"Dr Rajendra Prasad Government medical college Tanda Kanra Himachal Pradesh India 176001","correspondingAuthor":false,"prefix":"","firstName":"Amit","middleName":"","lastName":"Rana","suffix":""},{"id":590829988,"identity":"beec5801-0d11-476b-b5f5-58d6c954c462","order_by":2,"name":"Manpreet kaur","email":"","orcid":"","institution":"Post graduate institute of medical edication and research Chandigarh India 160012","correspondingAuthor":false,"prefix":"","firstName":"Manpreet","middleName":"","lastName":"kaur","suffix":""}],"badges":[],"createdAt":"2026-02-12 13:17:44","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8862488/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8862488/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":103165100,"identity":"73778a80-7c17-43ab-b0db-c6490161b556","added_by":"auto","created_at":"2026-02-22 12:25:01","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":187355,"visible":true,"origin":"","legend":"\u003cp\u003eCombined ROC curves for tumour asphericity (ASP) predicting four clinical outcomes. The dashed diagonal “Chance” line represents random classification (AUC = 0.50).\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8862488/v1/7e4db4062783869436077508.png"},{"id":103165171,"identity":"191984e2-80e7-4b49-a487-2de22bd8bcf2","added_by":"auto","created_at":"2026-02-22 12:25:04","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":252062,"visible":true,"origin":"","legend":"\u003cp\u003eComposite Forest plots illustrating the independent effect of high tumour asphericity (ASP \u0026gt; 0.25) alongside clinical and metabolic covariates across four multivariate models.Point estimates mark the adjusted odds ratio (in top panels) or hazard ratio (in bottom panels) for the listed covariates; Error bars: horizontal lines show the 95 % confidence intervals around each point estimate. Vertical dashed line indicates null effect (OR = 1 or HR = 1).\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8862488/v1/3b727bd928617cffa96b9c2d.png"},{"id":103165152,"identity":"2e4899b2-f206-4f36-9a57-d2689135ad93","added_by":"auto","created_at":"2026-02-22 12:25:04","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":215649,"visible":true,"origin":"","legend":"\u003cp\u003eThree panel depiction of tumour asphericity (ASP) and its relationship to baseline metastatic status, metabolic treatment response, and change in FDG uptake. \u003cstrong\u003e(A)\u003c/strong\u003e Boxplot of ASP for patients without (n = 130) versus with (n = 72) synchronous distant metastasis. Boxes denote median and interquartile range; whiskers extend to 1.5 × IQR.\u003cbr\u003e\n \u003cstrong\u003e(B)\u003c/strong\u003e Box plot of ASP for metabolic non responders (n = 50) versus responders (n = 107), defined by ≥ 30 % reduction in SUVₘₐₓ. \u003cstrong\u003e(C)\u003c/strong\u003e Scatter plot of individual ASP values versus percentage SUVₘₐₓ reduction on interim PET.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-8862488/v1/e08c11988a98d78c5df84811.png"},{"id":103165096,"identity":"bc4456db-66f3-40ca-ba3a-f78fe76939e6","added_by":"auto","created_at":"2026-02-22 12:24:59","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":200303,"visible":true,"origin":"","legend":"\u003cp\u003eKaplan-Meier survival curves stratified by tumour asphericity (ASP \u0026gt; 0.25 vs ≤ 0.25) for the overall cohort. Left: overall survival over 60 months. Right: progression free survival over 60 months. High asphericity is shown in yellow and low asphericity in red; log-rank p values are shown in each panel.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-8862488/v1/47e66c5ebf0d59284b1eabed.png"},{"id":103165113,"identity":"99af5785-333d-41f7-951f-7f43714503ea","added_by":"auto","created_at":"2026-02-22 12:25:03","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":245457,"visible":true,"origin":"","legend":"\u003cp\u003eKaplan-Meier survival curves stratified by tumour asphericity (ASP \u0026gt; 0.25 vs ≤ 0.25) within molecular subtypes. Top row: HR+/HER2−; middle row: HER2+; bottom row: triple-negative breast cancer. Left column: overall survival; right column: progression free survival. Log-rank p values are shown in each panel; high asphericity is shown in yellow and low asphericity in red.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-8862488/v1/9bfc469c8f44ef2f929b35ff.png"},{"id":103165172,"identity":"b35922c3-7c41-423d-bcb6-0a7ba5b79a9e","added_by":"auto","created_at":"2026-02-22 12:25:05","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":161551,"visible":true,"origin":"","legend":"\u003cp\u003eDecision curve analysis for 5 year progression free survival comparing net benefit of clinical, ASP, metabolic, and combined prediction strategies across threshold probabilities 0 – 0.50. The combined model provides the highest net benefit across most clinically relevant thresholds, relative to treat none and treat all reference strategies.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-8862488/v1/39cd39819d8a62e6f4d2cf60.png"},{"id":107870545,"identity":"95b4e949-4b8c-4b8d-94b9-3f6267783aae","added_by":"auto","created_at":"2026-04-27 07:39:54","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1354183,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8862488/v1/d6c71fb6-3195-40f7-ace1-ac93e0d60697.pdf"},{"id":103165173,"identity":"3ec4641f-6d0e-48ff-a6a1-7014ae871ab2","added_by":"auto","created_at":"2026-02-22 12:25:05","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":116175,"visible":true,"origin":"","legend":"","description":"","filename":"Image1.docx","url":"https://assets-eu.researchsquare.com/files/rs-8862488/v1/1b9375a35f70930937ab5bd3.docx"},{"id":103165099,"identity":"e1d1c9ab-c8e9-498f-9c11-2c6a9789786b","added_by":"auto","created_at":"2026-02-22 12:25:00","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":117466,"visible":true,"origin":"","legend":"","description":"","filename":"Image2.docx","url":"https://assets-eu.researchsquare.com/files/rs-8862488/v1/22134f2bfa1291afa7d3a7dc.docx"},{"id":103165103,"identity":"d2852674-87b1-4752-8ebf-61f3fa0cd827","added_by":"auto","created_at":"2026-02-22 12:25:01","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":84157,"visible":true,"origin":"","legend":"","description":"","filename":"Image3.docx","url":"https://assets-eu.researchsquare.com/files/rs-8862488/v1/0e7a944f78b3660296eae775.docx"}],"financialInterests":"","formattedTitle":"Primary tumour asphericity on baseline 18F-FDG PET/CT provides incremental prognostic value beyond stage and metabolic burden in breast carcinoma: association with metastasis, early metabolic response, and survival outcomes","fulltext":[{"header":"Introduction","content":"\u003cp\u003eBreast cancer is the most commonly diagnosed malignancy in women worldwide and remains a leading cause of cancer mortality[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Despite major advances in screening, systemic therapy, and multidisciplinary care, outcomes remain heterogeneous, ranging from long term cure in localized disease to early relapse and death in biologically aggressive or disseminated presentations [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eBecause prognosis and treatment selection depend strongly on stage and tumour biology, accurate baseline risk stratification is central to individualized treatment planning [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. However, routine clinicopathologic variables (tumour size, grade, receptor status, nodal involvement) may not fully capture spatial and metabolic complexity that could underlie early dissemination and variable response to therapy.\u003c/p\u003e \u003cp\u003eBaseline staging investigations for distant metastasis are generally not recommended for asymptomatic stage I-II disease but are considered for stage III because of the higher prevalence of occult metastatic spread and the potential for management changing findings [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Whole body \u003csup\u003e18\u003c/sup\u003eF-FDG PET/CT provides integrated anatomic and metabolic assessment and is increasingly used for intermediate and high risk breast cancer, where it can upstage disease, guide biopsy of equivocal lesions, and inform systemic and locoregional treatment strategies [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFor quantitative assessment, conventional PET biomarkers, SUVmax, metabolic tumour volume (MTV), and total lesion glycolysis (TLG), capture uptake intensity and tumour burden but only partially reflect intra tumoural spatial complexity and morphologic irregularity. Systematic reviews and meta-analyses support prognostic associations for MTV/TLG in breast cancer, but effect sizes vary across studies and meaningful overlap persists among patients with similar clinicopathologic features. Robust quantification also depends on harmonized acquisition and reconstruction, as emphasized by contemporary procedure guidelines [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eTreatment response on PET is commonly summarized using established metabolic response frameworks (e.g., the 1999 EORTC recommendations and PERCIST), which largely reduce complex spatial change to summary measures of uptake or lesion burden [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Consequently, there is interest in complementary descriptors that capture tumour geometry and heterogeneity and may improve risk stratification at baseline and during therapy.\u003c/p\u003e \u003cp\u003eRadiomics extends quantitative imaging by extracting high dimensional descriptors of intensity distribution, texture, and geometry that may act as imaging surrogates of tumour phenotype and microenvironment [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Standardization initiatives such as the Image Biomarker Standardisation Initiative (IBSI) provide consensus definitions and reporting guidance that can improve reproducibility across software implementations and institutions [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAmong radiomic feature families, shape metrics are attractive because they are intuitive and comparatively less dependent on intensity normalization. Asphericity, defined as 1 minus sphericity, quantifies deviation from an ideal sphere and can be derived from three dimensional lesion geometry. In FDG avid malignancies such as non small cell lung cancer and head and neck cancer, pretherapeutic asphericity has demonstrated independent prognostic value beyond standard metabolic indices [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. In breast cancer, PET derived radiomic and volumetric signatures (including texture features) have shown prognostic relevance, but the specific contribution of baseline primary tumour asphericity has not been systematically evaluated in routine clinical cohorts [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eConceptually, asphericity summarizes how \u0026ldquo;irregular\u0026rdquo; a metabolic tumour volume appears in three dimensions, integrating boundary complexity into a single scalar value. Because it is computed from the lesion\u0026rsquo;s surface area and volume, it can be implemented with modest computational overhead once segmentation is available. This practical profile makes ASP a candidate for inclusion in structured PET/CT reports alongside conventional metrics, provided segmentation and acquisition are standardized.\u003c/p\u003e \u003cp\u003eUsing a single institutional cohort of women with histologically confirmed breast carcinoma who underwent baseline \u003csup\u003e18\u003c/sup\u003eF-FDG PET/CT prior to systemic therapy, we tested whether primary tumour asphericity is associated with synchronous distant metastasis (primary endpoint) and overall survival, and secondarily with early metabolic non response and progression free survival, and whether it adds incremental prognostic value beyond clinical stage and conventional PET metrics [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eWhy this matters clinically: In routine practice, staging and treatment intensity in breast cancer are driven mainly by clinical stage and tumour biology, yet patients with similar stage and metabolic burden can diverge markedly in metastatic risk and outcomes. A simple shape descriptor derived from the primary tumour metabolic volume, such as asphericity, can help identify biologically aggressive disease with a higher likelihood of occult dissemination and early non response. If validated prospectively, reporting ASP alongside stage, subtype, and MTV/TLG could support risk adapted management, including triaging borderline stage II-III patients for comprehensive metastatic work up, guiding early escalation or switch of systemic therapy in predicted non responders, and tailoring surveillance and clinical trial enrolment.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003e\u003cstrong\u003eStudy Population\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis retrospective study was approved by the institutional review board with waiver of informed consent. Clinical records were screened to identify consecutive patients with newly diagnosed, biopsy confirmed breast carcinoma who underwent baseline \u003csup\u003e18\u003c/sup\u003eF-FDG PET/CT between January 2017 and December 2021 and received standard first line treatment (neoadjuvant chemotherapy [NAC] with or without targeted therapy, or upfront surgery). For patients receiving NAC, interim PET/CT was typically performed after two cycles of chemotherapy for early metabolic non response assessment; exact imaging intervals were not uniformly recorded. All patients had clinical follow up for survival endpoints.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eExclusion criteria:\u003c/strong\u003e Multifocal primary tumours, prior malignancy, incomplete imaging, or follow up \u0026lt; 6 months.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePET/CT Acquisition\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll studies were acquired on a GE Discovery IQ 610 (32 slice) hybrid PET/CT system (GE Healthcare, Waukesha, WI, USA). Patients received \u0026nbsp;~ 3.7 MBq/kg of \u003csup\u003e18\u003c/sup\u003eF-FDG, and imaging was performed after an uptake time of 60 \u0026plusmn; 10 minutes. Whole body PET emission data were acquired in 3D mode using a multi bed position protocol (2 minutes per bed position) with bed overlap as per institutional protocol. PET images used for analysis were reconstructed using Q.Clear (BSREM; DICOM reconstruction method QCHD) with CT based attenuation correction, on a 192 \u0026times; 192 matrix with voxel size 3.65 \u0026times; 3.65 \u0026times; 3.23 mm (reconstruction diameter 700 mm). Low dose CT for attenuation correction and anatomical localization was performed at 120 kVp with automatic tube current modulation, reconstructed at 512 \u0026times; 512 with 0.85 \u0026times; 0.85 mm in plane pixel spacing and 1.25 mm slice thickness (pitch 0.9375, rotation time 0.7 s; reconstruction diameter 434 mm). Multiplanar reformats (axial, coronal, sagittal) were generated on the workstation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImage Processing\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTumour segmentation, surface area, volume, and asphericity calculations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTumour segmentation and three dimensional (3D) geometric analysis were performed in Python (v3.10) using NiBabel, scikit-image, SciPy, and Trimesh. Whole body PET/CT studies were converted from DICOM to NIfTI (PET_converted.nii.gz) using dicom2nifti, preserving image orientation and voxel geometry. To reduce off-target seed placement and constrain the segmentation search space, segmentation was spatially constrained using a CT derived thoracic body mask (HU \u0026minus;500 to +1000; morphological closing radius 20 voxels) encompassing the thoracic field (including the breasts) while excluding abdominal and pelvic physiologic uptake. TotalSegmentator\u0026nbsp;(v2.12.0) [16] and nnU-Net (v1.7.1)\u0026nbsp;[17] were used to generate organ and axillary lymph node (ALN) exclusion masks to support accurate primary breast tumour delineation.\u003c/p\u003e\n\u003cp\u003eLesions were delineated using the scikit-image random walker algorithm with automated seed initialization. Foreground seeds were sampled from high-uptake voxels within the constrained PET volume using a foreground candidate threshold of 5% of SUVmax, while background seeds were sampled from low-uptake voxels using a 2% of SUVmax threshold (beta = 5; solver tolerance = 1\u0026times;10⁻\u0026sup2;). These thresholds were used only to define candidate seed pools for robust initialization and were not applied as intensity cut offs for the final tumour boundary. The final tumour label was exported as a binary mask and underwent post processing (morphological closing radius 2 voxels, hole filling for cavities \u0026ge;500 voxels, and boundary regularization) to ensure continuity and topological consistency. All post processing was applied only to the binary label mask and did not modify PET voxel intensities.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSurface mesh extraction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor surface based shape analysis, the tumour mask was resampled to an isotropic 1 mm grid using nearest neighbour interpolation (label preserving) and a triangular surface mesh was extracted using the Marching cubes algorithm (skimage.measure.marching_cubes). The resulting vertices and faces were imported into Trimesh to compute enclosed tumour volume (V, mm\u0026sup3;) and tumour surface area (A, mm\u0026sup2;). Tumour sphericity (\u0026Psi;) and asphericity (ASP) were calculated as: \u0026Psi; = \u0026pi;\u003csup\u003e\u0026nbsp;(1/3)\u0026nbsp;\u003c/sup\u003e(6V) \u003csup\u003e(2/3)\u003c/sup\u003e\u003csup\u003e\u0026nbsp;\u003c/sup\u003e/\u0026nbsp;A, and ASP = 1 - \u0026Psi;, where V is tumour volume (mm\u0026sup3;) and A is tumour surface area (mm\u0026sup2;).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eQuality control of tumour segmentation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eQC overlays were generated for every case by superimposing the binary contour on co-registered PET uptake in three orthogonal planes at the lesion centroid (Mask_QC.py; matplotlib, scikit-image) and saved as high resolution images. Two nuclear medicine physicians independently reviewed all overlays. Segmentations were flagged if the contour failed to enclose metabolically active tumour, included adjacent non tumour uptake, or missed obvious extensions/satellite foci.\u003c/p\u003e\n\u003cp\u003eA CT based thoracic body mask (HU -500 to +1000; morphological closing radius = 20 voxels) was applied before seeding to restrict seed placement to the thoracic field and avoid abdominal/pelvic physiologic uptake. When QC suggested misplaced or insufficient seeding, a manual seed coordinate (seed_xyz) was placed within the primary breast lesion and segmentation was repeated (\u0026lt;5% of cases). Mask voxel count and volume were logged; outliers were re-reviewed. Only masks passing visual and quantitative checks were included for mesh extraction and VOI metrics, and all QC overlays were archived. Reproducibility was assessed by repeat segmentation in a subset, and ASP agreement was summarized using the intraclass correlation coefficient (ICC). Interobserver ICC (intraclass correlation coefficient) for ASP was 0.94 (95% CI 0.90-0.97). \u0026nbsp;\u003cstrong\u003eImage 1 (from patient # 1) \u0026nbsp;\u003c/strong\u003eshow example of a left primary breast tumour segmetation from whole body FDG PET/CT scan DICOM images. \u003cstrong\u003eImage 2\u003c/strong\u003e (from patient # 2) shows segmentation of a right \u0026nbsp;primary breast tumour from whole body FDG PET/CT scan DICOM images in another patient. \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImage 3\u0026nbsp;\u003c/strong\u003eshows surface mesh 3D rendered orthogonal images of the same tumors from patient 1 and patient 2 respectively processed and generated with meshlab software (MeshLab 2025.07).\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eVOI Statistics (SUV, MTV, TLG)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo derive SUV based VOI metrics, raw PET values were converted to SUV units using DICOM metadata (Radiopharmaceutical Information Sequence) and patient weight: SUV(x,y,z) = (PETraw \u0026times; RescaleSlope + RescaleIntercept) \u0026times; Wg / DBq, where Wg is patient weight in grams and DBq is injected dose in Bq per scanner metadata. Importantly, VOI metabolic parameters were computed from the native resolution SUV PET image to preserve quantitative accuracy; PET intensities were not smoothed, filtered, or resampled for SUV calculations. To avoid misalignment effects, VOI masks used for quantification were maintained on the native (coarse) PET grid; where mapping between grids was required, masks were transferred using nearest neighbour resampling. PET derived quantitative metrics included SUVmax (maximum SUV voxel within the tumour VOI), SUVpeak (mean SUV within a fixed 1 cm\u0026sup3; spherical volume positioned to maximize local mean uptake within the tumour VOI), and SUVmean (mean SUV of all voxels within the tumour VOI). MTV (cm\u0026sup3;) was calculated as the number of tumour voxels multiplied by native PET voxel volume (mm\u0026sup3;) divided by 1000, and TLG (g\u0026middot;cm\u0026sup3;) was calculated as SUVmean \u0026times; MTV. All steps (conversion, segmentation, label processing, resampling for mesh, mesh metrics, and VOI metrics) were executed in a fully automated pipeline using fixed parameters applied uniformly across cases. To verify that mesh reconstruction did not influence SUV quantification, SUV metrics were derived exclusively from the native SUV image, whereas mesh derived measures were computed only from the label mask.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEndpoints\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eStudy endpoints comprised synchronous metastatic disease at presentation, early metabolic non response on interim PET/CT, overall survival (OS), and progression free survival (PFS). Synchronous (de novo) distant metastasis was defined as distant metastatic disease on baseline PET/CT, supported by correlative imaging and/or histopathology when available (histopathologic confirmation in n = 57); equivocal findings were adjudicated using follow up imaging and clinical documentation.\u003c/p\u003e\n\u003cp\u003eEarly metabolic response on interim PET/CT (performed for neoadjuvant chemotherapy patients where available, typically after two cycles) was defined a priori as a 30% reduction in primary tumour SUVmax relative to baseline, analogous to the PERCIST 1.0 partial metabolic response threshold but applied to SUVmax in this retrospective cohort. Early metabolic non response was coded as failure to achieve this \u0026ge; 30% SUVmax reduction. %DeltaSUVmax = [(SUVmax_baseline - SUVmax_interim) / SUVmax_baseline] x 100.\u003c/p\u003e\n\u003cp\u003eOverall survival (OS) was defined as the time from baseline PET/CT to death from any cause; patients were censored at last known follow up. PFS was defined as the time from baseline PET/CT to the earliest of (i) first documented disease progression or (ii) death from any cause, whichever occurred first; patients without an event were censored at last event free follow up. Follow-up imaging was not protocol mandated and was typically performed as PET/CT at clinical suspicion of recurrence. Accordingly, progression was determined primarily by the first PET/CT report describing unequivocal progressive disease (e.g., new FDG avid lesions consistent with metastasis and/or clear interval increase in extent or intensity of previously documented malignant uptake), with correlative imaging and/or histopathology used when available. For ROC analyses, OS and PFS were evaluated as 5 year endpoints (event within 5 years vs event free/censored beyond 5 years).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOptimal ASP cut off for metastasis detection was derived by ROC analysis (Youden index). The primary endpoint was synchronous distant metastasis at presentation (logistic regression); OS was the key time to event endpoint (Cox proportional hazards), with early metabolic non response and PFS analysed as secondary endpoints. Multivariable models adjusted for age, TNM stage, molecular subtype, tumour grade, and PET metabolic metrics (SUVmax, MTV, TLG). Collinearity among continuous predictors was assessed using variance inflation factors; ridge-penalized Cox modelling (L2) was used as a sensitivity analysis to obtain stable estimates when evaluating overlapping metabolic indices. Kaplan-Meier curves compared OS and PFS by ASP strata (log-rank test), and proportional hazards assumptions were checked using Schoenfeld residuals. Decision curve analysis for 5 year PFS compared net benefit of clinical-only, ASP only, metabolic only, and combined models across threshold probabilities 0 - 0.50 versus treat none and treat all strategies (apparent performance; no optimism correction). Statistical analyses were performed in MiniTab version 22.1(Minitab, LLC., PA, USA) and MedCalc (version 23.0.9, MedCalc Software Ltd, Ostend, Belgium) ; ridge sensitivity analyses were performed in Python. P value \u0026lt;0.05 was considered statistically significant.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003ePatient Characteristics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOf 245 screened patients, 202 met inclusion criteria. Median age was 52 years (range 28-81); 61% were hormone receptor positive/HER2-negative, 23% HER2-positive, and 16% triple negative. Median primary tumour SUVmax was 12.4 (IQR 8.7-17.9), and median ASP was 0.22. Interobserver intraclass correlation coefficient (ICC) for ASP was 0.94 (95% CI 0.90-0.97). Demographic, clinical, and metabolic metrics are shown in Table 1.\u003c/p\u003e\n\u003cp\u003ePrimary analyses focused on synchronous distant metastasis at presentation and overall survival; early metabolic non response and progression free survival were evaluated as secondary endpoints.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAssociation between TS, ASP and Baseline Metastasis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eROC analysis yielded an optimal TS threshold of 0.75 (equivalent ASP 0.25; AUC 0.78, 95% CI 0.71-0.84). Metastatic disease was present in 52% of high ASP (\u0026gt;0.25) versus 22% of low ASP tumours (p\u0026lt;0.001). In multivariable logistic regression, high ASP independently predicted metastasis (OR 3.1, p = 0.001) along with tumour size (OR 1.8 per cm, p = 0.02).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTumour ASP and Metabolic Treatment Response\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eInterim PET/CT was available for 157 patients receiving NAC. Early metabolic non response was observed in 32% of low ASP tumours versus 59% of high ASP tumours (p = 0.002). Baseline ASP inversely correlated with percentage SUVmax reduction (r = -0.50, p \u0026lt; 0.001).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSurvival outcomes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAfter median follow up of 48 months, 41 deaths and 78 progression events occurred. Five year OS was 88% for low ASP \u0026le; 0.25 vs 70% for high ASP \u0026gt; 0.25; corresponding five year PFS was 64 % vs 42 % respectively. High ASP remained an independent predictor of OS (HR = 2.4, p = 0.01); an association with PFS was also observed (HR = 1.9, p = 0.004) after adjustment for age, TNM stage, molecular subtype, tumour grade, SUVmax, MTV and TLG.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eROC Analysis\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe ROC panel (Figure 1) based on tumour asphericity (ASP; ASP = 1 - TS) demonstrated good discrimination across all pre-specified endpoints, with AUCs ranging from 0.78 to 0.81. Discrimination was strongest for 5 year progression free survival (AUC = 0.81) and early metabolic non response (failure to achieve \u0026ge;30% SUVmax reduction) (AUC = 0.80), followed by 5 year overall survival (AUC = 0.79) and synchronous baseline metastasis (AUC = 0.78). At a false positive rate of 0.20, sensitivities remained \u0026ge; 0.60 across endpoints, indicating that most events could be identified while limiting false positive classifications.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eComparison of tumour asphericity with clinical Stage, histopathology, grade and metabolic Burden\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eComprehensive multivariate models were constructed to benchmark tumour asphericity (ASP) against conventional prognosticators.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMetastatic disease at presentation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn an 8 variable logistic regression high ASP (\u0026gt; 0.25, corresponding to TS \u0026lt;0.75) retained the strongest independent association with distant metastasis (adjusted OR = 3.1, 95% CI 1.6-6.0, p = 0.001), exceeding the effect sizes of advanced clinical stage (OR = 2.2 for stage III; OR = 5.8 for stage IV) and metabolic volume (OR = 1.12 per 10 mL MTV).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEarly metabolic non response\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn multivariable logistic regression, baseline high asphericity (ASP \u0026gt; 0.25; TS \u0026lt; 0.75) was associated with a 2.5 fold higher odds of early metabolic non response (p = 0.002), independent of stage, subtype, grade, SUVmax, MTV and TLG.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eOverall and progression free survival\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn Cox models, ASP remained a significant predictor of OS (HR = 2.4, p = 0.01); an association with PFS was also observed (HR = 1.9, p = 0.004) after simultaneous adjustment for all covariates, ranking second only to stage IV disease.\u003c/p\u003e\n\u003cp\u003eTumour asphericity (ASP) and its relationship to baseline metastatic status, metabolic treatment response, and change in FDG uptake are shown as box plots and scatter plots in Figure 3. Kaplan-Meier survival curves stratified by tumour asphericity (ASP \u0026gt; 0.25 vs \u0026le; 0.25) for breast carcinoma in general and for different subtypes are shown in Figure 4 and Figure 5, indicating statistically significant differences in survival between subgroups. These comparisons underscore that morphometric heterogeneity captured by ASP offers additive prognostic information beyond tumour burden and molecular phenotype.\u003c/p\u003e\n\u003cp\u003eDecision curve analysis for 5 year PFS (Figure 6) showed greater net benefit for ASP only and metabolic only models compared with the clinical only model, while the combined model (clinical + ASP + metabolic metrics) provided the highest net benefit across most clinically relevant thresholds (0 - 0.50) relative to treat none and treat all strategies.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIndependent determinants across endpoints\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMultivariable models confirmed that morphometric, staging, histologic, and metabolic variables each contributed independent prognostic information. Figure 2 shows Forest plots illustrating the independent effect of high tumour asphericity (ASP \u0026gt; 0.25) alongside clinical and metabolic covariates across four multivariate models.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMetastasis:\u003c/strong\u003e High tumour ASP (\u0026gt; 0.25) and stage IV disease showed the largest independent effects; triple negative phenotype and higher grade were also associated with metastatic presentation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEarly metabolic non response:\u0026nbsp;\u003c/strong\u003eBaseline high asphericity (ASP \u0026gt; 0.25; TS \u0026lt; 0.75) increased the odds of non response (failure to achieve \u0026ge;30% SUVmax reduction) after adjustment for stage, subtype, grade and metabolic indices.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eOverall survival\u003c/strong\u003e: High ASP (\u0026gt; 0.25), stage IV disease and triple negative subtype were associated with inferior OS, with additional contributions from baseline metabolic burden.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eProgression free survival\u003c/strong\u003e (secondary endpoint): Similar covariates predicted earlier progression or death, with stage IV disease and ASP showing the largest effect sizes.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this retrospective cohort, baseline \u003csup\u003e18\u003c/sup\u003eF-PET/CT derived primary tumour asphericity (ASP) was independently associated with synchronous distant metastasis and inferior overall survival, even after adjustment for established clinicopathologic variables and conventional metabolic biomarkers (SUVmax, MTV, TLG). Associations with early metabolic non response and progression free survival were also observed, although PFS ascertainment was clinically driven. Overall, these results support ASP as a simple, interpretable morphometric biomarker that captures complementary information beyond uptake intensity and volumetric burden.\u003c/p\u003e \u003cp\u003eA key implication is that tumour geometry on baseline \u003csup\u003e18\u003c/sup\u003eF-FDG PET/CT may capture aspects of aggressiveness that are not fully represented by intensity or volume metrics alone. MTV and TLG reflect tumour burden and have been linked to prognosis in breast cancer, but they do not explicitly quantify the spatial organization of uptake within the tumour volume [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. ASP, by contrast, operationalizes geometric irregularity and may summarize complex metabolic contours associated with infiltrative growth, stromal reaction, or heterogeneous viable tumour distribution.\u003c/p\u003e \u003cp\u003eOur findings align with prior work in other cancers. In primary non small cell lung cancer and head and neck squamous cell carcinoma, asphericity of pretherapeutic FDG uptake stratified survival independently of tumour volume and uptake intensity [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Together with these reports, our results support the broader concept that shape based descriptors can provide clinically relevant information across tumour types, even when derived from routine clinical imaging protocols.\u003c/p\u003e \u003cp\u003eBreast cancer PET radiomics literature has predominantly emphasized texture and intensity based features, which can reflect metabolic heterogeneity and have been associated with recurrence risk or survival in selected cohorts [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. By focusing on a geometric descriptor that is computationally straightforward, our work complements existing approaches and may be easier to communicate clinically. From an interpretability standpoint, ASP can be described as a single measure of how far the metabolic tumour departs from a sphere, which may support clinician acceptance compared with high dimensional texture signatures.\u003c/p\u003e \u003cp\u003eThe biological interpretation of ASP is necessarily indirect. A more irregular uptake geometry could reflect a mixture of viable tumour, necrosis, stromal reaction, and heterogeneous perfusion, with uptake contours shaped by local microenvironmental constraints and infiltrative tumour fronts. Hypoxia is a recognized driver of invasion and metastatic dissemination in breast cancer, and hypoxia regulated programs can remodel extracellular matrix and promote tumour cell survival during dissemination [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. These mechanisms align with broader frameworks describing how cancers acquire invasive and metastatic capabilities [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Radiomics studies across tumour types further suggest that macroscopic imaging phenotypes can encode aspects of underlying biology and clinical behaviour, providing a rationale for why even simple geometry based features may remain informative after adjustment for intensity and volume metrics [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFrom a clinical decision standpoint, ASP may help refine which patients are most likely to benefit from comprehensive staging and intensified surveillance. Practice guidance discourages routine baseline metastatic imaging for asymptomatic early stage disease but supports consideration for stage III presentations; within such a framework, ASP could further triage intermediate and high risk patients for targeted work up, confirmatory imaging, or closer follow up when conventional risk indicators yield borderline or discordant signals [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eASP may also complement established PET response frameworks. While EORTC and PERCIST emphasize changes in uptake intensity (and, in practice, lesion burden), geometry based descriptors may capture a different dimension of response, such as irregular residual uptake or fragmentation, that could be relevant when uptake changes are modest, heterogeneous, or confounded by inflammatory uptake [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Prospective studies could explore combined response models integrating uptake, volumetrics, and geometry over time, and evaluate whether ASP improves early identification of non responders or patients at risk of early progression.\u003c/p\u003e \u003cp\u003eTechnical and implementation considerations merit attention. ASP is derived from surface area and volume of a segmented lesion; therefore, it can be sensitive to segmentation strategy, voxel size, smoothing, and partial volume effects, particularly in small lesions or lesions with low contrast. However, because ASP is a shape based metric derived from surface topology, it is theoretically less susceptible to inter scanner SUV calibration errors than intensity based metrics like SUVmax or TLG, provided the segmentation boundary remains topologically consistent. Harmonized acquisition and reconstruction, standardized SUV calibration, and transparent segmentation protocols are essential to reduce measurement variability in PET metrics and derived radiomic features [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Standardized feature definitions and reporting guidance, as promoted by IBSI, further support reproducibility across software implementations and centres [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. In practice, adopting a limited set of well defined, clinically interpretable features (e.g., MTV, TLG, ASP) may provide a pragmatic bridge between conventional reporting and more complex radiomic signatures [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFrom a modelling perspective, improvements in discrimination and net benefit when ASP was added to multivariable models suggest potential utility across clinically plausible decision thresholds. Decision curve analysis is particularly valuable for assessing whether a model improves decision making by balancing harms of false positive actions against missed high risk patients [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. In this study, decision curve analysis was performed for 5 year PFS (a secondary endpoint) to illustrate potential downstream impact on escalation of staging, follow up, or therapy.\u003c/p\u003e \u003cp\u003eFuture work should focus on multicentre validation with prespecified workflows that include automated or semi-automated segmentation, harmonized acquisition and reconstruction, and prospective evaluation of whether ASP improves clinical decision making when embedded in structured reports. Such studies should also quantify test-retest repeatability and assess sensitivity to segmentation strategy and voxel geometry so that practical reporting thresholds can be defined for routine use [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e].\u003c/p\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003eLimitations of the study\u003c/h2\u003e \u003cp\u003eLimitations of the study include its retrospective nature, single centre design introduces potential selection bias and limits generalizability. ASP may vary with delineation strategy, lesion size, and imaging protocol; although we used consistent acquisition and segmentation, external validation should assess robustness across scanners, reconstruction settings, and segmentation paradigms [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. In addition, the study was not designed to link ASP directly to histopathologic correlates such as tumour architecture, necrosis, or hypoxia markers, and such correlative studies would strengthen mechanistic interpretation. Finally, longer follow up and larger cohorts are needed particularly for hormone receptor positive disease where late recurrences are common and treatment decisions are nuanced [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusions","content":"\u003cp\u003eTumour asphericity extracted from baseline \u003csup\u003e18\u003c/sup\u003eF-FDG PET/CT offers an easily obtainable, reproducible biomarker that stratifies risk beyond conventional factors. In our cohort, high asphericity (\u0026gt;\u0026thinsp;0.25) independently predicted synchronous metastasis and inferior OS after controlling for stage, molecular subtype, grade and metabolic tumour burden; associations with early metabolic non response and PFS were supportive secondary findings. Integration of this simple shape metric into routine PET/CT reports could refine personalised decision making, identifying patients who may benefit from intensified staging, systemic therapy, closer surveillance or inclusion in clinical trials. Prospective multicentre validation and evaluation of dynamic changes during therapy are warranted.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e: This retrospective study was approved by the institutional review board with a waiver of informed consent.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication:\u003c/strong\u003e Waived due to the retrospective nature of the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;Funding\u003c/strong\u003e: None.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests:\u003c/strong\u003e The authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eUse of AI assisted technologies\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAn AI language model was used to assist with language editing and clarity of the manuscript. All authors reviewed and edited the content and take full responsibility for the accuracy and integrity of the work. No AI tool was used to generate, alter, or analyze patient data, statistical results, or clinical images.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCode availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCustom Python code used for tumour segmentation, mesh extraction, and asphericity computation will be made available in a public repository (e.g., GitHub/Zenodo) at the time of publication; during peer review it can be shared with editors and reviewers on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics statement: This retrospective study was approved by the institutional review board with a waiver of informed consent.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePatient consent for publication: Waived due to the retrospective nature of the study.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding: None.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of interest: None declared.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions declarations :\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDr. Nitin Gupta (M.B;B.S. M.D. nuclear medicine , Medical Image Analysis (IIT Madras (NPTEL), Certificate in Advanced AI in imaging (RSNA, USA):\u0026nbsp;\u003c/strong\u003eConceptualization; Methodology; Software (Python based tumour segmentation and asphericity pipeline); Formal analysis; statistics, visualization; writin \u0026nbsp;original draft; review \u0026amp; editing; Critical revision of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDr. Amit Rana, M.D. ; D.M Medical Oncology:\u0026nbsp;\u003c/strong\u003eLiterature review; \u0026nbsp;manuscript review, editing; Manuscript formatting.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMs. Manpreet Kaur, M.Sc. R.S.O and medical physicist (Nuclear medicine) :\u0026nbsp;\u003c/strong\u003eData curation (retrieving patient DICOM data \u0026nbsp;and images and clinical data); Patient follow up and data collection.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAll authors read and approved the final manuscript and agree to be accountable for all aspects of the work.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate:\u0026nbsp;\u003c/strong\u003eThis retrospective study was approved by the Institutional Ethics Committee (IEC), Dr. Rajendra Prasad Government Medical College, Kangra (Tanda), Himachal Pradesh, India, with a waiver of informed consent due to the retrospective design and use of anonymized data.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eSung H, Ferlay J, Siegel RL, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209-249. doi:10.3322/caac.21660. PMID:33538338.\u003c/li\u003e\n\u003cli\u003eWaks AG, Winer EP. Breast cancer treatment: a review. N Engl J Med. 2019;381(3):288-300. doi:10.1056/NEJMra1807166. PMID:31483975.\u003c/li\u003e\n\u003cli\u003eArnaout A, Varela NP, Allarakhia M, et al. Baseline staging imaging for distant metastasis in women with stages I, II, and III breast cancer. Curr Oncol. 2020;27(2):e123-e145. doi:10.3747/co.27.6147. PMID:32489262.\u003c/li\u003e\n\u003cli\u003eGroheux D, Hindie E, Delord M, et al. 18F-FDG PET/CT for staging and restaging of breast cancer. J Nucl Med. 2016;57(Suppl 1):17S-26S. doi:10.2967/jnumed.115.157859. PMID:26834096.\u003c/li\u003e\n\u003cli\u003eBoellaard R, Delgado-Bolton R, Oyen WJG, et al. FDG PET/CT: EANM procedure guidelines for tumour imaging: version 2.0. Eur J Nucl Med Mol Imaging. 2015;42(2):328-354. doi:10.1007/s00259-014-2961-x. PMID:25452219.\u003c/li\u003e\n\u003cli\u003eWen W, Xuan D, Hu Y, et al. Prognostic value of maximum standard uptake value, metabolic tumor volume, and total lesion glycolysis of positron emission tomography/computed tomography in patients with breast cancer: A systematic review and meta-analysis. PLoS One. 2019;14(12):e0225959. doi:10.1371/journal.pone.0225959. PMID:31826010.\u003c/li\u003e\n\u003cli\u003eYoung H, Baum R, Cremerius U, et al. Measurement of clinical and subclinical tumour response using [18F]-fluorodeoxyglucose and positron emission tomography: review and 1999 EORTC recommendations. Eur J Cancer. 1999;35(13):1773-1782. doi:10.1016/S0959-8049(99)00229-4. PMID:10673991.\u003c/li\u003e\n\u003cli\u003eWahl RL, Jacene H, Kasamon Y, Lodge MA. From RECIST to PERCIST: Evolving considerations for PET response criteria in solid tumors. J Nucl Med. 2009;50(Suppl 1):122S-150S. doi:10.2967/jnumed.108.057307. PMID:19403881.\u003c/li\u003e\n\u003cli\u003eLambin P, Leijenaar RTH, Deist TM, et al. Radiomics: the bridge between medical imaging and personalised medicine. Nat Rev Clin Oncol. 2017;14(12):749-762. doi:10.1038/nrclinonc.2017.141. PMID:28975929.\u003c/li\u003e\n\u003cli\u003eAerts HJWL, Velazquez ER, Leijenaar RTH, et al. Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat Commun. 2014;5:4006. doi:10.1038/ncomms5006. PMID:24892406.\u003c/li\u003e\n\u003cli\u003eHa S, Choi H, Paeng JC, Cheon GJ. Radiomics in oncological PET/CT: a methodological overview. Nucl Med Mol Imaging. 2019;53(1):14-29. doi:10.1007/s13139-019-00571-4. PMID:30828395.\u003c/li\u003e\n\u003cli\u003eZwanenburg A, Valli\u0026egrave;res M, Abdalah MA, et al. The Image Biomarker Standardisation Initiative: standardized quantitative radiomics for high-throughput image-based phenotyping. Radiology. 2020;295(2):328-338. doi:10.1148/radiol.2020191145. PMID:32154773.\u003c/li\u003e\n\u003cli\u003eApostolova I, Rogasch J, Steffen IG, et al. Asphericity of pretherapeutic tumor FDG uptake in primary non-small cell lung cancer: a new prognostic marker. BMC Cancer. 2014;14:896. doi:10.1186/1471-2407-14-896. PMID:25444154.\u003c/li\u003e\n\u003cli\u003eApostolova I, Steffen IG, Wedel F, et al. Asphericity of pretherapeutic tumor FDG uptake provides independent prognostic value in head-and-neck cancer. Eur Radiol. 2014;24(9):2077-2087. doi:10.1007/s00330-014-3269-8. PMID:24965509.\u003c/li\u003e\n\u003cli\u003eGroheux D, Martineau A, Teixeira L, et al. 18F-FDG PET/CT for predicting the outcome in ER+/HER2- breast cancer patients: comparison of clinicopathological parameters and PET image-derived indices including tumor texture analysis. Breast Cancer Res. 2017;19:3. doi:10.1186/s13058-016-0793-2. PMID:28057031.\u003c/li\u003e\n\u003cli\u003eWasserthal J, Breit HC, Meyer MT, Pradella M, Hinck D, Sauter AW, et al. \u003cstrong\u003eTotalSegmentator: Robust segmentation of 104 anatomic structures in CT images.\u003c/strong\u003e \u003cstrong\u003eRadiol Artif Intell.\u003c/strong\u003e 2023;5(5):e230024. doi:10.1148/ryai.230024. PMID:37795137.\u003c/li\u003e\n\u003cli\u003eIsensee F, Jaeger PF, Kohl SAA, Petersen J, Maier-Hein KH. \u003cstrong\u003ennU-Net: a self-configuring method for deep learning-based biomedical image segmentation.\u003c/strong\u003e \u003cem\u003eNature Methods.\u003c/em\u003e 2021;18(2):203-211. doi:10.1038/s41592-020-01008-z. PMID:33288961. \u003c/li\u003e\n\u003cli\u003e18. Semenza GL. Molecular mechanisms mediating metastasis of hypoxic breast cancer cells. Trends Mol Med. 2012;18(9):534-543. doi:10.1016/j.molmed.2012.08.001. PMID:22921864.\u003c/li\u003e\n\u003cli\u003eHanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144(5):646-674. doi:10.1016/j.cell.2011.02.013. PMID:21376230.\u003c/li\u003e\n\u003cli\u003e. Vickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models. Med Decis Making. 2006;26(6):565-574. doi:10.1177/0272989X06295361. PMID:17099194.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Table","content":"\u003cp\u003e\u003cstrong\u003eTable 1.\u0026nbsp;\u003c/strong\u003eShowing demographics, imaging \u0026amp; survival metrics of patients included in the study.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eSection\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eCategory\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003en\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e%\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eMedian SUVmax (IQR)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eMedian MTV, mL (IQR)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eMedian TLG, g (IQR)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eMedian ASP (IQR)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eMedian PFS (months)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eMedian OS (months)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge group\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eAge 20-29\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e2.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e9.4 (7.2-12.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e14.6 (8.2-20.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e129.6 (70.4-180.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.29 (0.23-0.30)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e38\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e53\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge group\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eAge 30-39\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e26\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e13%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e11.2(9.8-12.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e15.8 (9.4-22.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e136.8 (80.3-190.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.26 (0.19-0.28)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e45\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e60\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge group\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eAge 40-49\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e56\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e27.7%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e10.6 (8.6-15.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e15.2 (10.3-23.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e147.2 (90.6-200.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.24 (0.19-0.29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e60\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge group\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eAge 50-59\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e34.7%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e12.8 (9.6-16.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e17.3 (11.6-25.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e168.2 (100.4-224.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.25 (0.20-0.30)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e58\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge group\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eAge 60-69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e17.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e11.7 (8.6-15.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e19.3 (12.4-28.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e173.6 (116.2-232.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.26 (0.21-0.31)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e38\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e52\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge group\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eAge \u0026ge;70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e5.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e13.6 (11.2-17.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e20.8 (13.6-30.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e180.6(120.9-240.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.28 (0.25-0.32)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e32\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e45\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eTNM stage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eStage I\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e5.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e8.4 (5.7-11.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e10.7(5.4-15.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e86.7 (40.7-121.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.18 (0.12-0.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e70\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eTNM stage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eStage II\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e42.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e10.6 (7.2-14.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e15.3 (8.7-24.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e123.6 (62.8-206.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.22 (0.17-0.27)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e48\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e60\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eTNM stage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eStage III\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e32.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e13.8(10.5-18.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e22.8 (12.7-34.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e180.7 (100.4-280.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.30 (0.26-0.35)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e50\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eTNM stage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eStage IV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e20.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e17.6 (14.2-21.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e28.1 (15.6-38.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e250.8 (150.4-350.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.33 (0.29-0.38)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e35\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eHistopath type\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eInvasive ductal carcinoma (NST)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e162\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e80.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e12.8 (8.7-16.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e17.9 (9.6-27.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e165.7 (82.3-250.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.26\u003c/p\u003e\n \u003cp\u003e(0.19-0.29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e55\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eHistopath type\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eInvasive lobular carcinoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e5.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e7.2 (5.4-9.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e14.2 (8.5-22.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e108.3(50.7-160.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.24 (0.15-0.25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e45\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e60\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eHistopath type\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eMucinous carcinoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e2.5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e6.4 (4.1-7.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e10.2 (5.7-18.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e80.7 (40.6-144.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.21 (0.13-0.24)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e65\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eHistopath type\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eMedullary carcinoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e11.6 (10.3-18.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e20.4 (12.7-30.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e184.2 (90.8-282.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.28 (0.23-0.33)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e38\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e48\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eHistopath type\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eMetaplastic / other special types\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e2.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e15.3 (12.1-21.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e24.3 (14.6-35.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e210.8 (110.6-328.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.30 (0.25-0.35)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e40\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eHistopath type\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eOthers (papillary, tubular, etc.)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e5.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e10.3 (7.2-14.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e15.2 (8.3-24.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e120.7 (60.8-204.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.22 (0.17-0.27)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e58\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eTumour grade\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eGrade I\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e14.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e7.8 (5.3-11.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e12.4 (6.2-20.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e84.9 (44.2-137.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.18 (0.12-0.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e65\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eTumour grade\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eGrade II\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e44.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e11.4(8.7-16.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e19.6 (11.2-30.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e163.8 (80.6-250.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.22 (0.17-0.27)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e56\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eTumour grade\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eGrade III\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e82\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e40.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e15.6 (11.8-21.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e23.5 (15.8-37.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e214.3 (116.5-327.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.28 (0.23-0.33)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e48\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e"},{"header":"Images","content":"\u003cp\u003eImages 1 to 3 are available in the Supplementary Files section.\u003c/p\u003e\n"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Breast cancer, 18F-FDG PET/CT, radiomics, tumour asphericity, metastasis, survival","lastPublishedDoi":"10.21203/rs.3.rs-8862488/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8862488/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eTumour sphericity (TS) and its inverse, tumour asphericity (ASP)\u0026thinsp;=\u0026thinsp;1 - TS, derived from baseline \u003csup\u003e18\u003c/sup\u003eF-FDG PET/CT may capture morphologic heterogeneity linked to tumour aggressiveness.\u003c/p\u003e\u003ch2\u003eAim\u003c/h2\u003e \u003cp\u003eTo determine whether baseline primary tumour asphericity on \u003csup\u003e18\u003c/sup\u003eF-FDG PET/CT provides incremental prognostic value beyond clinical stage, molecular subtype, histologic grade, and PET metabolic burden (SUVmax, MTV, TLG) for synchronous distant metastasis and survival outcomes in breast carcinoma.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eThis retrospective study included 202 women (median age 52 years) with biopsy-confirmed breast cancer staged by \u003csup\u003e18\u003c/sup\u003eF-FDG PET/CT (January 2017 to December 2021) and followed for a median of 48 months. Primary endpoints were synchronous distant metastasis at presentation and overall survival (OS); early metabolic non response and progression-free survival (PFS) were secondary endpoints. Primary tumours were semi automatically segmented (Python; Trimesh) and TS/ASP were computed from 3D surface geometry. Optimal cut offs were identified by ROC analysis. Early metabolic non response was defined as failure to achieve\u0026thinsp;\u0026ge;\u0026thinsp;30% reduction in primary tumour SUVmax on interim PET/CT (where available). Multivariable logistic and Cox models were adjusted for age, TNM stage, molecular subtype, tumour grade, and PET metabolic metrics (SUVmax, MTV, TLG).\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eMedian TS was 0.78 (IQR 0.72\u0026ndash;0.84); an ASP threshold of 0.25 optimally separated metastatic from non-metastatic cases (AUC 0.78, 95% CI 0.71\u0026ndash;0.84). High ASP (\u0026gt;\u0026thinsp;0.25) independently predicted synchronous distant metastasis (OR 3.1, 95% CI 1.6-6.0; p\u0026thinsp;=\u0026thinsp;0.001) and was associated with inferior 5 year OS after adjustment (HR 2.4, 95% CI 1.2\u0026ndash;4.7; p\u0026thinsp;=\u0026thinsp;0.010). High ASP (\u0026gt;\u0026thinsp;0.25; TS\u0026thinsp;\u0026lt;\u0026thinsp;0.75) increased the likelihood of early metabolic non response (OR 2.5, 95% CI 1.4\u0026ndash;4.5; p\u0026thinsp;=\u0026thinsp;0.002). Associations with PFS were also observed (HR 1.9, 95% CI 1.2\u0026ndash;2.9; p\u0026thinsp;=\u0026thinsp;0.004). In decision curve analysis for 5 year PFS, models including ASP showed higher net benefit than clinical only models across clinically relevant thresholds.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eBaseline tumour asphericity is a reproducible, interpretable biomarker of metastatic dissemination and adverse survival in breast cancer, complementing stage, tumour biology, and metabolic burden. Its association with early metabolic non response supports a role for PET/CT based risk stratification and treatment tailoring, pending multicentre validation.\u003c/p\u003e","manuscriptTitle":"Primary tumour asphericity on baseline 18F-FDG PET/CT provides incremental prognostic value beyond stage and metabolic burden in breast carcinoma: association with metastasis, early metabolic response, and survival outcomes","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-22 12:24:30","doi":"10.21203/rs.3.rs-8862488/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"7a534198-291d-4697-b072-7bdd8e6666da","owner":[],"postedDate":"February 22nd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-27T05:41:55+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-22 12:24:30","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8862488","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8862488","identity":"rs-8862488","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}