Topological Analysis of the Human Lymph Node Reticular Network Predicts Outcome in Breast Cancer

Topological Analysis of the Human Lymph Node Reticular Network Predicts Outcome in Breast Cancer | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (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];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Topological Analysis of the Human Lymph Node Reticular Network Predicts Outcome in Breast Cancer Amy Llewellyn , Sharon D’Costa , Ringo Lam , Jasmine Gore , Veronika Lachina , Daniel Shewring , View ORCID Profile Sophie E. Acton , View ORCID Profile Kalnisha Naidoo doi: https://doi.org/10.1101/2025.07.07.663481 Amy Llewellyn 1 Translational Pathology, Comprehensive Cancer Centre, School of Cancer and Pharmaceutical Sciences, King’s College London , London, UK Find this author on Google Scholar Find this author on PubMed Search for this author on this site Sharon D’Costa 2 Synnovis, Department of Cellular Pathology , a partnership between SYNLAB UK & Ireland , Guy’s and St Thomas’ NHS Foundation Trust, and King’s College Hospital NHS Foundation Trust , London, UK Find this author on Google Scholar Find this author on PubMed Search for this author on this site Ringo Lam 2 Synnovis, Department of Cellular Pathology , a partnership between SYNLAB UK & Ireland , Guy’s and St Thomas’ NHS Foundation Trust, and King’s College Hospital NHS Foundation Trust , London, UK Find this author on Google Scholar Find this author on PubMed Search for this author on this site Jasmine Gore 1 Translational Pathology, Comprehensive Cancer Centre, School of Cancer and Pharmaceutical Sciences, King’s College London , London, UK Find this author on Google Scholar Find this author on PubMed Search for this author on this site Veronika Lachina 3 Stromal Immunology Group, Laboratory for Molecular Cell Biology, University College London , London, UK Find this author on Google Scholar Find this author on PubMed Search for this author on this site Daniel Shewring 3 Stromal Immunology Group, Laboratory for Molecular Cell Biology, University College London , London, UK Find this author on Google Scholar Find this author on PubMed Search for this author on this site Sophie E. Acton 3 Stromal Immunology Group, Laboratory for Molecular Cell Biology, University College London , London, UK Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Sophie E. Acton Kalnisha Naidoo 1 Translational Pathology, Comprehensive Cancer Centre, School of Cancer and Pharmaceutical Sciences, King’s College London , London, UK 4 Department of Cellular Pathology, King’s College Hospital , London, UK Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Kalnisha Naidoo For correspondence: kalnisha.1.naidoo{at}kcl.ac.uk Abstract Full Text Info/History Metrics Supplementary material Preview PDF SUMMARY Axillary lymph nodes (ALN) initiate local immune responses in breast cancer (BC) but how and when ALN become dysfunctional, facilitating metastasis, is unclear. We use unbiased computational approaches to quantify features of ALN stromal architecture. We identify PDGFRβ as a robust, immunomarker for human fibroblastic reticular cells (FRC) and use it to quantify how FRC network topology changes during BC progression and after treatment. ALN (n = 331) from 179 BC patients and 23 benign reactive controls were assessed for FRC network metrics, including lacunarity and branchpoints, alongside de-identified clinico-pathological data. We find in node-negative, triple-negative BC, neoadjuvant treatment induced denser FRC networks which correlated with improved survival. Conversely, denser FRC networks in node-positive patients correlated with worsened survival, regardless of BC subtype or treatment. Further, increased FRC alignment within metastases improved survival. We show that FRC network topology predicts prognosis in BC, providing a new avenue for mechanistic, translational research. INTRODUCTION In breast cancer (BC), axillary lymph node (ALN) status is central to guiding management decisions since sentinel lymph node (SLN) involvement is the most significant prognostic factor in early-stage disease. However, while ALN are the first site of BC metastasis, they are also highly organized, tightly regulated secondary lymphoid organs capable of generating powerful adaptive immune responses against tumor cells. Furthermore, ALN modulate fluid homeostasis in the arm and breast. Consequently, surgical clearance of the axilla can result in lymphoedema of the arm but the effects on immune function are still unclear 1 . Over the past 30 years, BC treatment has evolved significantly, now integrating tailored combinations of neoadjuvant and adjuvant systemic therapies based on disease stage and hormone receptor status (estrogen receptor positive (ER+), human epidermal growth factor receptor 2 positive (HER2) and triple negative breast cancer (TNBC)). In parallel, management of the axilla has also changed, with the number of ALN clearances (ALNC) decreasing in favor of less invasive techniques such as SLN biopsy (SLNB) and, more recently, targeted axillary dissection 2 . Clinical trials, such as Z0011 and AMAROS, have led to this paradigm shift by challenging the necessity of ALNC in patients with minimal axillary involvement 3 , 4 . In particular, the fact that isolated tumor cells (ITCs; < 200 cells or < 0.2 mm in extent) and micrometastasis (0.2–2 mm in extent) do not impact recurrence or overall disease-free survival, if appropriate systemic therapies are given, supports conserving the axilla 5 . Despite this however, precise histopathological examination of every surgically removed ALN is still mandated to ensure accurate prognostication and staging. Furthermore, with neoadjuvant immunotherapy now being standard of care for TNBC, the potential of ALN microenvironment to either activate or suppress the immune response is clinically relevant 6 . However, our biological understanding of the mechanisms of ALN metastasis, and the factors that determine if a patient will develop an effective anti-tumor immune response or not, remains poorly understood. Fibroblastic reticular cells (FRC) are integral to ALN structure and function. These specialized stromal cells secrete and enwrap extracellular matrix to form a network of reticular fibers that facilitate efficient immune cell localization and antigen transport (reviewed in 7 ). Lymph enters the lymph node via the subcapsular sinus (SCS), where sinus-lining cells act as a molecular sieve, restricting molecules larger than ∼70 kDa from accessing the conduit system 8 – 10 . Smaller soluble molecules filter into the FRC ensheathed conduit network and flow in a unidirectional, controlled manner through the lymph node parenchyma to high endothelial venules (HEV). In addition to their structural scaffolding role, FRC orchestrate immune cell interactions through producing chemokines such as chemokine ligand (CCL)19 and CCL21 which regulate lymphocyte migration. Single cell RNA sequencing (scRNAseq) has elucidated distinct subsets of murine FRC, each with highly specific roles in maintaining immunological niches 11 . However, the canonical marker of murine FRC remains Podoplanin (Pdpn). In mice, this membranous glycoprotein has been shown to have central role in the maintenance and dynamic remodeling of the conduit network, and in immunoregulatory properties of FRC 12 – 14 . In contrast, our knowledge of human FRC biology and their response to immune activation and cancer is still evolving. Studying human FRC is challenging due to the ethical constraints on accessing lymphoid tissue. In BC patients, every excised ALN must be formalin-fixed and paraffin-wax embedded (FFPE) and examined in its entirety for accurate pathological staging 15 . This limits the number of patient-derived ALN samples available for research. In addition, FRC make up only 1-5% of human lymph nodes and are significantly more difficult to isolate from primary tissue than immune cells 16 , 17 . Furthermore, it is widely recognized that subclassifying human fibroblasts is challenging due to the lack of unique, robust markers (reviewed in 18 ). Attempts to isolate human FRC have traditionally used PDPN to define the population, but PDPN expression is low in human FRC 19 – 21 . However, very recent scRNAseq of reactive human lymph nodes has revealed that platelet-derived growth factor receptor β (PDGFRβ) is expressed by all subsets of lymph node fibroblasts 22 . Finally, the collagenous reticular network formed by murine FRC also exists in human lymph nodes, yet our understanding of how its structure and composition change in disease remains limited. Histopathologists have historically used tinctorial stains such as reticulin to visualize this delicate network 23 and Masson’s trichrome to highlight mature fibrosis 24 , but these methods have not been correlated with functional and/or clinical data, or FRC topography. There are limited data on freshly resected ALN containing BC metastases, which identify four fibroblast subsets with prognostic significance based on the differential expression of five fibroblast markers: fibroblast activation protein α1 (FAP), integrin β1, smooth muscle actin (SMA), PDGFRβ, and PDPN 25 . However, this study was limited by a small sample size, the absence of benign reactive controls and did not define the spatial distribution of these FRC populations within ALN. Herein, we define robust immunohistochemical (IHC) markers of human FRC and characterize their expression patterns in ALN from a large cohort of BC patients, as well as reactive, control, tumor-free lymph nodes from patients with benign disease. We comprehensively map and quantify the effect of BC and neoadjuvant chemotherapy (NACT) on FRC network topology in both involved and uninvolved ALN, linking changes in FRC network topology to patient survival. RESULTS Optimized IHC Markers of Human FRC Subsets We analyzed 331 FFPE ALN samples from 179 BC patients diagnosed and treated at King’s College Hospital (KCH), encompassing all three molecular subtypes, as well as different TNM stages and variable responses to NACT ( Table 1 ). No patients had received checkpoint inhibitor therapy (CPI). 23 tumor-free lymph nodes from patients with benign disease, who had never had cancer, served as reactive controls. Given the paucity of robust IHC stains for human lymph node FRC, we optimized protocols for five previously reported FRC markers 26 and compared staining patterns across serial sections to reticulin, a classic tinctorial stain which is known to highlight the collagen backbone of lymphoreticular organs 23 . View this table: View inline View popup Table 1. Clinico-pathological Characteristics of BC Patient Cohort PDGFRβ mirrored reticulin staining, proving to be the most robust pan-FRC marker. It showed consistently strong staining in T cell zones, weak follicular dendritic cell (FDC) staining in germinal centers, and strong capsular fibroblast staining ( Figure 1 ). SMA strongly marked FRC in the paracortex and capsule, with weak staining in parafollicular zones; it was absent in germinal centers. PDPN, strongly stained lymphatic endothelial cells and moderately labelled germinal center FDC, but did not highlight T zone FRC. Integrin β1 was restricted to mature blood vessels and the lymph node capsule. FAP staining proved technically challenging despite extensive optimization. The high temperatures required to retrieve antigen and achieve a signal damaged the tissue, and therefore FAP was not pursued further. Download figure Open in new tab Figure 1. Optimized IHC Markers of Human FRC Subsets These representative photomicrographs show optimized IHC staining of reactive human lymph node tissue from patients with benign disease (n = 23 nodes; scale bars, 50 µm; all images at x20 magnification). (A-C) The reticular network was not visible by H&E staining in the germinal centers (A), paracortex (T cell zone; B) or capsule (C). (D-F) Reticulin staining highlighted the reticular collagen network weakly in germinal centers (D) but strongly in the paracortex (E). The capsule (F) demonstrated a distinct reticulin staining pattern, characterized by thicker, wavier fibers. (G-I) Masson’s trichrome did not stain the FRC network in germinal centers (G) or the T cell zone (H), but stained the mature, capsular collagen blue (I). (K-L) PDGFRβ stained FRC in the paracortex (K), with weak FDC staining in germinal centers (J) and strong fibroblast staining in the capsule (L). This pattern mirrored reticulin staining (D-F). (M-O) SMA was absent in germinal centers (M), but stained FRC in the paracortex (N) and the capsule (O). (P-R) PDPN stained FDC in germinal centers (P) and lymphatic endothelial cells (R), but did not label FRC in the T cell zone (Q). (T-U) Integrin β1 staining was restricted to mature blood vessels and the capsule (U), with no staining of FDC (S) or FRC (T). Quantitative Metrics Capture FRC Network Architecture in Human Lymph Nodes Whole slide images were scanned and representative 250,000 µm² regions of interest (ROI) were chosen from PDGFRβ-stained T cell zones. ROI were chosen from reactive nodes from patients with benign disease, uninvolved nodes from patients with BC, residual lymphoid tissue from nodes with smaller metastatic tumor deposits (< 2 mm in maximal extent), areas of NACT-induced fibrosis and areas of metastatic tumor ( Figure 2 ). Since subtle alterations in FRC topology could potentially have large implications for tissue function, we needed a precise, unbiased approach to quantify changes in response to BC and/or NACT. The Workflow Of Matrix BioLogy Informatics (TWOMBLI) is an image analysis pipeline designed to quantify a broad range of parameters in the extracellular matrix in normal and pathological tissue 27 . Due to the intimate association between FRC and the reticulin network, we were able to repurpose this pipeline to analyze FRC network parameters 7 . TWOMBLI analysis showed that the FRC network in reactive lymph nodes (benign controls; clinico-pathological characteristics shown in Supplementary Table 1 ) has low lacunarity (median = 6, interquartile range (IQR) = 5-7.8), moderate branching (median branchpoints normalized by fiber length = 0.066, IQR = 0.059-0.068), consistent FRC size (median width = 1.5 µm, IQR = 1.3-2.0); median fiber length = 24 µm, IQR 23-34) and are poorly aligned (median alignment = 0.053, IQR = 0.04-0.077). Interestingly, these network features were not significantly correlated with patient age ( Supplementary Table 2 ); however, this finding should be interpreted with caution due to the small sample size and the use of reactive nodes, which may not accurately reflect the uninflamed aging process. Download figure Open in new tab Figure 2. Quantitative Metrics Capture FRC Network Architecture in Human Lymph Nodes ROI were selected for TWOMBLI analysis from reactive benign control (n = 23); uninvolved (n = 214); tumor-infiltrated (n = 75); and fibrotic (n = 16) nodes, as well as residual lymphoid tissue in metastatic nodes (n = 72). (A, F, K and P) The reticular network was not apparent on H&E-stained images (bird’s eye view; scale bars, 1mm). High-magnification (x20) images of boxed areas, each an example of analyzed ROI, are shown in B, G, I, L and Q (scale bars, 50 µm). (C, H, M and R) PDGFRβ highlighted a dense FRC network in reactive benign control and uninvolved nodes (C), as well as in residual lymphoid tissue in nodes containing small tumor deposits that were at least 500µm away (H; images at x20 magnification; scale bars, 50 µm). (J and R) In metastatic nodes, the small (J) and large (R) tumor deposits showed a disrupted, stretched FRC network (images at x20 magnification; scale bars, 50 µm). (M) In lymph nodes where tumor had been eradicated by neoadjuvant chemotherapy, leaving residual fibrosis, PDGFRβ highlighted only a few remaining, aligned FRC within this area (image at x20 magnification; scale bar, 50 µm). (D, N and S) From each PDGFRβ-stained ROI, TWOMBLI masks were generated to extract matrix features. This highlighted increased lacunarity in ALN containing large tumor deposits (S) compared to reactive nodes (D). (E, O and T) TWOMBLI masks were then processed and thresholded to calculate HDMI. This showed reduced HDMI in post-NACT fibrosis (O) compared to reactive nodes (E). Treatment-Naïve TNBC and NACT Remodel the FRC Network in Uninvolved ALN Multivariate linear analysis of each TWOMBLI output from uninvolved nodes removed from BC patients against the clinico-pathological variables ( Table 1 ) revealed statistically significant associations ( Supplementary Table 3 ). Number of endpoints in the reticular network yielded the highest R score (0.127, p = 0.000007), with molecular subtype (p = 0.0006) and tumor burden in the axilla (p = 0.01; explained below) as significant predictors. Fiber width (p = 0.00006), number of branchpoints (p = 0.00009), lacunarity (p ≤ 0.0001), fiber length (p = 0.00007), high-density matrix intensity (HDMI; p = 0.001) and hyphal growth unit (HGU; p = 0.05) were all significantly predicted by NACT exposure. To assess network parameters alongside clinical variables in a more integrative manner, we performed a Principal Component Analysis (PCA) on the combined dataset ( Figure 3A-C ). The first principal component (PC1) accounted for 26.09% of the total variance, predominantly driven by lacunarity (loading = -0.506), followed by branchpoints (loading = 0.478), and NACT (loading = 0.403) ( Supplementary Table 4 and Supplementary Figure 1 ). This suggests that PC1 captures variability driven by both network structural features and clinical context. Other PCs incorporated additional clinical and network parameters, indicating complex multi-variable relationships across topological and clinical parameters ( Supplementary Table 4 ). Strikingly, the benign control lymph nodes always clustered separately from uninvolved ALN from BC patients, suggesting cancer-specific changes to ALN structure. Download figure Open in new tab Figure 3. Treatment-Naïve TNBC and NACT Remodel the FRC Network in Uninvolved ALN (A-C) PCA was performed to integrate the TWOMBLI derived FRC network features and clinico-pathological variables obtained from reactive, benign control nodes (n = 23) and uninvolved ALN (ER+: n = 40; HER2: n = 78; and TNBC: n = 96). Each data point represents the median of 8 ROI from each node. PC1 accounted for 26.09% of the total variance, predominantly driven by lacunarity (loading = -0.506), followed by branchpoints (loading = 0.478), and then NACT (loading = 0.403). PC2 and PC3 contributed 16.34% and 15.72% respectively to variance. See also Figure S1 and Table S2. Benign, reactive control nodes always clustered separately to uninvolved ALN from BC patients. (D-G) Violin plots showing molecular subtype-specific differences in FRC network lacunarity (D), HDMI (E), number of branchpoints (F) and number of endpoints (G) in uninvolved ALN *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001 (Kruskal-Wallis test). Median with interquartile range, with minimum and maximum. See also Figure S2. (H) PCA showing effect of NACT on uninvolved nodes from all BC molecular subtypes (treatment naïve: n = 126 ALN; NACT: n = 111 ALN). Each data point represents the median of the results from 8 ROI from each node. (I, J and K) Violin plots illustrating the impact of NACT on FRC network lacunarity (I), number of branchpoints (J) and number of endpoints (K) in uninvolved ALN, stratified by BC molecular subtype (reactive: n = 23 nodes, HER2 treatment-naïve: n = 39 nodes, HER2 post-NACT: n = 39 nodes, TNBC treatment-naïve: n = 29 nodes, TNBC post-NACT: n = 67 nodes). Each data point represents the median of 8 ROI from each node. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001 (Kruskal-Wallis test). Median with interquartile range, with minimum and maximum. (L and M) Kaplan–Meier survival analysis of TNBC patients with ypN0 disease (i.e. no residual disease in the axilla post-NACT) shows that patients with uninvolved ALN that show a lacunarity ≥ 5.7 (p = 0.0365) or branchpoints <0.065 (p = 0.0051) have a significantly worse overall survival (n = 49 nodes; Gehan-Breslow-Wilcoxon method). Further, we observed topological alterations linked to BC subtype. Stratification of TWOMBLI results from both treatment naïve and post-NACT patients by BC molecular subtype revealed that FRC networks in uninvolved nodes from patients with TNBC had significantly reduced lacunarity (p ≤ 0.001; Figure 3D ), unchanged HDMI ( Figure 3E ), increased branchpoints (p = 0.0048; Figure 3F ) and fewer endpoints (p ≤ 0.0001; Figure 3G ) compared to HER2 and reactive nodes, independently of treatment. Visualization of the PCA distribution by NACT exposure also revealed a clear segregation between treatment-naïve and NACT-exposed patients ( Figure 3H ). Subsequent faceting of the PCA by both molecular subtype and NACT exposure further demonstrated distinct separation between uninvolved and reactive control nodes ( Supplementary Figure 2 ). However, the separation was more pronounced in TNBC than HER2 positive disease and was even greater in nodes from patients who had received NACT, indicating that both tumor subtype and therapy exposure contribute to network remodeling. Further stratification of individual TWOMBLI parameters according to subtype and NACT exposure confirmed these patterns. FRC networks from NACT-exposed TNBC patients displayed significantly decreased lacunarity (p ≤ 0.01; Figure 3I ), increased branchpoints (p ≤ 0.01; Figure 3J ) and decreased endpoints (p ≤ 0.0001; Figure 3K ; Supplementary Figure 2) . Collectively, these findings indicate that in treatment-naive TNBC patients, the FRC network in uninvolved ALN is more compact and highly branched. Interestingly, exposure to NACT induces comparable alterations throughout the ALN basin. Notably, Kaplan–Meier analysis of patients with TNBC and no axillary nodal involvement following NACT (pathological nodal stage 0, ypN0) demonstrated that the presence of a denser FRC network, characterized by lower lacunarity (p = 0.0365, Figure 3L ) and a higher number of branchpoints (p = 0.0051, Figure 3M ), correlated with improved survival. Metastatic Tumor Deposits Drive Greater FRC Network Disruption Within Residual Nodes In order to test the effect of BC on the FRC network within a single node as well as across the ALN chain, we repeated our analysis including ROI of areas of residual lymphoid tissue in involved nodes from patients of all molecular subtypes, both treatment naïve and post-NACT. Multivariate analysis revealed lacunarity as the parameter most significantly influenced by clinical factors (R score 0.157, p = 2×10⁻⁹) with the size of metastasis (i.e. ITCs; micrometastasis; or macrometastasis), NACT exposure and molecular subtype as the most significant predictors ( Supplementary Table 5 ). Direct comparison of TWOMBLI outputs from treatment-naive and post-NACT nodes confirmed the presence of a denser, more branched network demonstrated in the uninvolved nodes following NACT ( Supplementary Figure 3 ). Furthermore, stratification by molecular subtype also confirmed a significant effect of TNBC on topology. PCA again demonstrated that across all subtypes, reactive control nodes from benign patients formed distinct clusters compared to BC patient-derived ALN ( Figure 4A-I ). Moreover, within each molecular subtype, metastatic ALN containing residual lymphoid tissue clustered separately to uninvolved nodes. This indicates that the effects of BC on the stromal network are more pronounced in the residual lymphoid tissue that sits adjacent to tumor deposits in metastatic ALN. Interestingly, lacunarity had the highest loading on PC1 (0.410), which accounted for 26.62% of the variance in the data ( Supplementary Table 6 and Supplementary Figure 1 ). This aligns with the results of the multivariate analysis ( Supplementary Table 5) . Furthermore, PC1, PC2, and PC3 all contributed to meaningful separation of the data ( Figure 4J-L ). Taken together, these results confirm that the presence of TNBC and exposure to NACT induce a more compact FRC network in the residual lymphoid tissue of involved nodes. Download figure Open in new tab Figure 4. Metastatic Tumor Deposits Drive Greater FRC Network Disruption Within Residual Nodes PCA, stratified by BC molecular subtype, was performed to integrate TWOMBLI derived FRC network features and clinico-pathological variables from reactive benign nodes (n = 23), uninvolved ALN and residual lymph node tissue from involved ALN. PC1, PC2, and PC3 contribute 26.62%, 14.85%, and 11.47% of the variance, respectively. Each data point represents the median of 8 ROI per node. See also Figure S1, Figure S3 and Table S4. Once again, reactive benign control nodes clustered distinctly from BC patient-derived nodes. (A, B and C) ALN from patients with ER positive BC (ER+ uninvolved: n = 40 nodes, ER+ residual: n = 29 nodes). (D, E and F) ALN from patients with HER2 positive disease (HER2 uninvolved: n = 78 nodes, HER2 residual: n = 15 nodes). (G, H and I) ALN from patients with TNBC (TNBC uninvolved: n = 96 nodes, TNBC residual: n = 28 nodes). (J, K and L) Loading plots illustrating the contribution of the top six individual features to the first three principal components. (M and N) Kaplan–Meier survival analysis of patients with pN1 or ypN1 disease, of any molecular subtype, showed that patients with uninvolved ALN with a lacunarity <5.5 have a significantly worse overall survival (p = 0.0407) but that branchpoints ≥ 0.065 (p = 0.271) is not a significant predictor (n = 82 nodes; Gehan-Breslow-Wilcoxon method). We specifically investigated the impact of tumor metastasis in ALN on the FRC topology of uninvolved nodes within the same chain and linked these data with patient survival. In patients with metastatic BC involving one to three lymph nodes (pathological nodal stage 1, pN1 (treatment-naïve) and/or ypN1 (post-NACT)), a more compact FRC network, indicated by reduced lacunarity, in the uninvolved nodes was significantly associated with poorer overall survival, independent of treatment status (p = 0.0407; Figure 4M ). However, in this group of patients, number of branchpoints was not correlated with survival (p = 0.271; Figure 4N ). Chemotherapy-Induced Fibrosis Replaces Reticular Network Architecture In reactive, benign control lymph nodes the delicate reticulin/FRC network spans the entire node ( Figures 5A-D ). In contrast, NACT produces wedge-shaped areas of fibrosis in which tumor cells are replaced by dense eosinophilic collagen and angiogenic vessels ( Figure 5E ) . This fibrosis clinically indicates a response to NACT 28 . However, the nature of this fibrosis and its effect on the FRC network have not been previously characterized. Reticulin staining of these areas revealed abundant mature collagen composed of thicker, wavy, argyrophilic fibers that replaced the normal reticular network ( Figure 5F ). These fibers were similar to those found within the lymph node capsule and also showed blue staining with Masson’s trichrome ( Figure 5G ), indicating the presence of mature collagen. PDGFRβ immunostaining demonstrated complete destruction of the network with fewer, scattered remaining FRC ( Figure 5H ). TWOMBLI analysis revealed increased FRC alignment (p = 0.0210; Figure 5M ), non-significant changes in HDMI ( Figure 5N ) and a significant reduction in FRC length (p = 0.002; Figure 5O ) in these fibrotic areas. However, TWOMBLI is not optimized for assessing alignment, particularly in areas where the network has been destroyed. Therefore, the same ROIs were additionally analyzed using CurveAlign 29 , which showed a trend towards increased FRC alignment (p = 0.09; Figure 5P ) in areas of post-NACT fibrosis when compared to matched, uninvolved ALN, but this did not reach statistical significance. Download figure Open in new tab Figure 5. Chemotherapy-Induced Fibrosis Replaces Reticular Network Architecture (A-D) Representative photomicrographs showing ROI from reactive, benign control lymph nodes (n = 23; x20 magnification, scale bars, 50 µm). The reticular network was not visible by H&E staining (A), but was highlighted by reticulin staining (B). Masson’s Trichrome (C) showed no mature collagen in the reactive lymph node parenchyma. PDGFRβ stained the delicate FRC network (D). (E-H) Representative photomicrographs showing ROI from areas of chemotherapy-induced fibrosis (n = 16 nodes; x20 magnification, scale bars, 50 µm). H&E staining (E) showed dense eosinophilic collagen with sparse lymphocytes. Reticulin staining (F) showed complete replacement of the normal reticular network by thicker argyrophilic fibers. Masson’s trichrome staining (G) demonstrated abundant mature (blue) collagen. PDGFRβ immunohistochemistry (H) revealed complete destruction of the FRC network with few remaining FRC. (I-L) Representative photomicrographs showing ROI from areas of uninvolved ALN that were removed from the same patients as those containing NACT-induced fibrosis (n = 16 nodes; x20 magnification, scale bars, 50 µm). H&E staining (I) showed only lymphoid tissue, with no areas of fibrosis and no evidence of tumor deposits. Reticulin staining (J) showed a normal reticular network. Masson’s trichrome stain (K) showed no mature collagen. PDGFRβ immunostaining (L) revealed a preserved FRC network. (M-P) Violin plots comparing FRC alignment (M and P), HDMI (N) and length (µm) across reactive lymph nodes (n = 23) to areas of chemotherapy-induced fibrosis and matched uninvolved (negative) ALN from the same patient (n = 16). FRC alignment calculated using both TWOMBLI software (M) and CurveAlign software (P;see methods for details). Each data point represents a median of the results from 4 ROI from each node. *p ≤ 0.05 (Kruskal-Wallis test). Median with interquartile range, with minimum and maximum. (Q) Representative photomicrograph of a PDGFRβ-stained ROI showing the interface between chemotherapy-induced fibrosis and the residual FRC network (n = 13 nodes; x20 magnification, scale bars, 50 µm). (R and S) Examples of heatmaps of TWOMBLI analysis generated from ROI in (Q) illustrating the gradual increase in HDMI (R) and decrease in lymphocyte cell number (S) across the transition from an area of residual lymph node to an area of NACT-induced fibrosis. (T and U) Violin plots of the aggregated heatmap data comparing HDMI (T) and number of lymphocytes (U) across areas of residual lymph node, interface and areas of NACT-induced fibrosis (n = 13 nodes). ***p ≤ 0.001, ****p <0.0001 (Kruskal-Wallis test). Median with interquartile range, with minimum and maximum. PDGFRβ staining at the interface between chemotherapy-induced fibrosis and the remaining FRC network showed the gradual transition from destroyed to disrupted to intact FRC network architecture ( Figure 5Q ). For high-resolution quantification, each ROI at the interface was subdivided into 100 tiles, which were individually analyzed using TWOMBLI. These tiles were then reconstructed into heatmaps to visually represent the gradual changes in network parameters across the residual node, interface and fibrotic area (examples shown in Figures 5R and 5S ). Aggregated heatmap data showed the highest average HDMI values in the interface region (p ≤ 0.0001; Figure 5T ), and a gradual decrease in cell (lymphocyte) number from the residual node through the interface to the fibrotic zone (p ≤ 0.0001; Figure 5U ). FRC Network Topology Correlates with Tumor Burden and Clinical Outcome in Metastatic ALN As anticipated, survival analysis of this cohort demonstrated a poorer prognosis for patients with a higher nodal stage ( Figure 6A ), as well as those with TNBC and a higher tumor stage ( Supplementary Figure 4 ). However, traditional nodal stage incorporates only the number of positive nodes and not the size of deposits, which we have shown to be highly relevant to network topology. Furthermore, in the clinical setting, axillary tumor burden is routinely quantified only in post-NACT samples and not in treatment naïve disease 30 . In order to compare these two patient cohorts, we adapted the formula used to calculate the residual cancer burden score to estimate axillary tumor burden (see Methods ) 30 . Stratification using this metric revealed that patients with a tumor volume ≥ 20 had significantly worse outcomes (p = 0.0075; Figure 6B ). As tumor cells invade ALN, they progressively distort the FRC network, quantified by a significant increase in lacunarity correlated with a decrease in HDMI ( Figure 6C-G ). We observed that the degree of this network distortion is correlated to the size of the metastatic deposit, rather than molecular subtype or NACT exposure ( Figure 6C-G ). Notably, large nests of tumor resulting in high network lacunarity (≥ 12) was associated with significantly poorer survival (p = 0.043, Figure 6H ). More surprisingly, in patients with BC of all molecular subtypes and treatment status, increased FRC alignment in metastatic deposits improved survival (p = 0.0205, Figure 6I-K ). Download figure Open in new tab Figure 6. FRC Network Topology Correlates with Tumor Burden and Clinical Outcome in Metastatic ALN (A and B) Kaplan–Meier survival curves showed reduced survival in patients with higher nodal stage (A; p = 0.0006) and axillary tumor volume ≥ 20 (B;p = 0.0075; n = 179 patients; Gehan-Breslow-Wilcoxon method). See also Figure S4. (C and D) Representative photomicrographs showing ROI of macrometastases with different lacunarities in ALN stained with PDGFRβ (x20 magnification, scale bar, 50 µm). (E) Scatterplot showing the inverse correlation between TWOMBLI derived lacunarity and HDMI in metastatic ALN. Data point color reflects metastatic tumor diameter (µm); larger metastases are associated with higher lacunarity and reduced FRC network density (n = 75 nodes; values for lacunarity and HDMI represent a median of the results from 8 ROI in each node). (F and G) Violin plots comparing FRC network HDMI (F) and alignment (G) in involved ALN, stratified according to lacunarity (involved ALN with lacunarity <12: n = 52; involved ALN with lacunarity ≥12: n = 23) to reactive benign control nodes (n = 23). Each data point represents a median of the results from 8 ROI from each node. **p ≤ 0.01, ****p ≤ 0.0001 (Kruskal-Wallis test). Median with interquartile range, with minimum and maximum. (H) Kaplan–Meier survival analysis, stratified by lacunarity in metastatic nodes, showed that patients with positive nodes showing high lacunarity have significantly poorer survival compared to those with low lacunarity (p = 0.043; n = 64 patients; Gehan-Breslow-Wilcoxon method). (I and J) Representative photomicrographs showing ROI of a macrometastasis with an alignment of 0.055 (I, calculated using CurveAlign software), as opposed to one with an alignment of 0.4 (J), in ALN stained with PDGFRβ (x20 magnification, scale bar, 50 µm). (K) Kaplan–Meier survival analysis, stratified by alignment in metastatic nodes, showed that patients with involved nodes showing an FRC alignment <0.2 (calculated using CurveAlign software) have significantly poorer survival compared to those with higher FRC alignment (p = 0.0205; n = 64 patients; Gehan-Breslow-Wilcoxon method). DISCUSSION By optimizing robust IHC markers for human FRC, we have shown how different subsets of FRC localize within the ALN of BC patients. We confirmed PDGFRβ as a pan-FRC marker that highlights FRC in all of the microanatomical ALN compartments, where different immune populations reside and traffic in both health and disease 22 . Interestingly, our data show that, in contrast to murine lymph nodes, where it stains FRC in the T cell zones, human PDPN is highly expressed predominantly by germinal center FDC and lymphatic endothelial cells 14 . In addition, SMA expression is more prominent in human tissue, where it is seen primarily in the paracortex, than in mice, where it typically stains weakly in germinal centers, increasing only upon activation. It is possible that this difference in staining patterns, rather than being an inherent variation between species, is age-related, since most mouse models use young, immunologically naïve mice, whereas in humans, even ‘normal’ nodes will have undergone repeated cycles of antigen exposure, immune activation and resolution before being removed and examined. Moreover, PDPN expression has been observed in FRC within tertiary lymphoid structures in human non-small cell lung cancer, indicating that its expression may be context-dependent 31 . Regardless, this biological variablity could compromise on-going efforts at clinical translational, and as such, should be investigated further.2 Because of the intimate association between FRC and collagen in lymph nodes 32 , we were able to repurpose TWOMBLI as an unbiased and quantitative tool to map the human FRC network in different pathophysiological conditions 27 . By integrating TWOMBLI derived parameters with clinical data through multivariate analysis, we were able to clearly visualize FRC network changes in a translationally relevant context. Although we focused on BC, our approach is applicable to other malignancies and pathologies. We have demonstrated that the FRC network becomes denser and more highly branched in treatment naïve TNBC, an aggressive, hard-to-treat BC subtype. We postulate that the denser FRC network seen in these patients may restrict the flow of lymph through the reticular network in the ALN parenchyma, shunting fluid through the SCS and into the efferent vessels. This change in fluid flow could likely have two clinically relevant effects. Firstly, the pooling of fluid in the SCS may provide an opportunity for cancer cells to extravasate and accumulate, leading to the development of metastatic tumor deposits 33 . Secondly, controlled flow through the lymph node parenchyma is critical for antigen sampling by tissue resident antigen-presenting cells. Without these interactions the ability of the node to generate effective immune responses against tumor antigens and pathogens would be compromised 10 . Our data show these changes trend towards a poorer prognosis in patients with treatment naïve TNBC, but this result did not reach statistical significance. This is most likely due to the fact that it is hard to collect tissue from treatment naïve patients, since most TNBC patients receive NACT at present 34 . Interestingly, these topological changes were evident in both patients without nodal metastasis and in uninvolved ALN from patients with metastatic deposits elsewhere in the axillary chain. Previous mouse studies have shown that factors secreted by primary melanoma travel to the draining lymph nodes, promoting FRC proliferation and a shift toward a cancer-associated fibroblast (CAF)-like phenotype 35 . Thus, it stands to reason that soluble factors originating in the primary breast tumor could be responsible for the changes seen in patients with pN0 disease. On the other hand, the presence of similar topological changes in uninvolved nodes from patients with nodal metastasis in other ALN (pN1/ ypN1) could be due to inter-nodal communication within the axillary chain through soluble factors. This makes it difficult to unpick precisely what is occuring in patients who received NACT, since these patients could have had metastatic disease in their axillary nodes which were eradicated before surgery. In other words, inter-nodal communication cannot be entirely excluded in ypN0 cases. This confounding effect is not seen once metastases are established in the axillary chain, however. In the uninvolved nodes from patients with pN1 or ypN1 disease of all subtypes, a reduction in lacunarity was associated with poorer prognosis, regardless of treatment status. This is clinically significant as treatment decisions regarding the extent of axillary surgery in the node-positive cohort are often challenging 3 . Therefore, any biomarker that can predict whether SLN have switched from ‘helping’ (immunocompetent) to ‘harming’ (immune-exhausted) a patient would be invaluable. Biologically, TNBC is thought to be more immunogenic than other BC subtypes. A subset of TNBC tumors are lymphocyte-predominant (i.e. contain tumor infiltrating lymphocytes amounting to at least 30% of the tumor) and ALN of patients with TNBC more often contain a greater number of GC than other BC subtypes 36 , 37 . Despite this, CPI are only effective if NACT is given concurrently to augment the anti-tumor immune response 38 . Nevertheless, only 50% of TNBC patients achieve a complete pathological response after NACT, and only a further 10% benefit from the addition of a CPI to neoadjuvant treatment regimens 39 . We have shown that the FRC network of uninvolved ALN is significantly denser and more highly branched post-NACT, and that this was more pronounced in TNBC patients. Intriguingly, in contrast to the treatment naïve patients, this change in FRC topology was associated with an improved survival in TNBC patients with no residual metastasis in the axilla. This suggests that the same alterations in fluid flow, with shunting into the SCS, may be beneficial in this cohort, potentially by facilitating drug-induced killing of small subcapsular tumor deposits. In metastatic ALN where chemotherapy eradicates large tumor deposits, it concurrently obliterates the FRC network, replacing it with dense, mature fibrotic collagen structures containing a few, highly aligned FRC and very occasional lymphocytes. These structural changes could have important functional consequences for impeding immune cell trafficking and activation, and this in turn might help to resolve whether patients who have a complete pathological response after NACT should have a targeted, rather than a complete, axillary dissection 40 . Moreover, since the addition of CPI to NACT results in mixed responses, understanding all of the factors that could influence treatment response is crucial. We aim to confirm and further characterize this in a larger cohort of patients in the future. Simulations indicate that the FRC network remains functionally resilient despite the loss of up to 50% of its structure 41 . We have shown that even though the FRC network is maintained as BC invades into a node, its topology is highly distorted. Furthermore, remodeling of the FRC network to a denser, more branched topology was observed in residual lymphoid tissue that was situated > 500 µm from metastatic deposits within a node. These changes were similar to those seen in the uninvolved nodes but to a greater extent. Simple diffusion of the tumor-derived soluble factors through a whole lymph node is unlikely to be responsible for these changes as the distances are too great 42 , 43 . It seems more plausible that the tumor-derived soluble factors driving these intranodal changes are transported through the altered conduit network. As axillary tumor burden increased in all of the BC patients in our cohort, survival significantly decreased. This reinforces the prognostic value of both the number of nodal deposits and the size of each deposit, in both the treatment naïve and post-NACT context. Furthermore, we have shown that the extent of FRC network distortion within the metastatic deposit correlates with the size of the deposit rather than the BC subtype or other clinical features, suggesting that physical disruption is the primary driver. Once a tumor deposit exceeds a lacunarity of 12, survival worsens significantly. This size threshold could indicate a critical point at which the altered flow of fluid through the FRC network compromises the ability of the node to mount effective responses, thereby adversely affecting survival. Conversely, increased FRC alignment in tumor deposits is associated with a better prognosis, which may reflect more controlled flow through the node parenchyma. These hypotheses warrant further investigation using in vivo and in silico models to fully elucidate the mechanisms by which FRC network architecture influences fluid flow and immune function. We intend to use the REPLICANT model, a unique ex vivo perfusion system for human lymph nodes, to study fluid flow in real time and identify key molecular drivers and/or clinically relevant biomarkers 44 – 46 . We have shown, that changes in FRC topology can be quantified using unbiased computational approaches in human ALN using IHC, and that these changes predict prognosis in BC. By integrating this with real time flow dynamics, we are now positioned to mechanistically disentangle these complex interactions in patient-derived samples in translationally relevant context in future. Limitations of the study None of the patients in this cohort received CPI, as these therapies were only recently introduced into routine clinical practice and therefore, outcome data for these samples would amount to less than 5 years 6 . Given the growing use of immunotherapy, particularly in TNBC, future studies should explore how combination therapies involving CPI and NACT impact FRC network architecture and function. In addition, although this was a longitudinal study with ALN tissue samples collected from patients at various stages of disease and with follow-up data extending up to 10 years, the availability of sequential samples from individual patients was limited. Despite this, we were able to identify meaningful changes in the FRC network over time. Future prospective studies with longitudinal sampling after surgical resection of the primary tumor would enhance our understanding of the permanence or potential reversibility of these stromal changes. RESOURCE AVAILABILITY Lead contact Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Kalnisha Naidoo ( kalnisha.1.naidoo{at}kcl.ac.uk ). Materials availability This study did not generate new unique reagents. Data and code availability All data reported in this paper will be shared by the lead contact upon request. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. AUTHOR CONTRIBUTIONS Conceptualization, K.N., S.E.A. and A.L.; methodology, A.L., S.D.C. R.L, J.G., V.L. and D.S.; investigation, A.L.; writing—original draft, A.L.; writing—review & editing, K.N. and S.E.A.; funding acquisition, A.L., K.N. and S.E.A.; resources, K.N., S.E.A., A.L., S.D.C and R.L.; supervision, K.N. and S.E.A. DECLARATION OF INTERESTS The authors declare no competing interests. SUPPLEMENTAL INFORMATION Document S1. Figures S1 – S4 and Tables S1 – S6. STAR*METHODS KEY RESOURCES TABLE View this table: View inline View popup EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS Patient Cohort Archival, FFPE human tissue samples were obtained through the Breast Cancer Immune, Drug and Gene (BRIDGE) Study (Research Ethics Committee No: 24/NW/0079). KCH pathology databases were searched to identify patients who had undergone a breast surgical excision, with either a SLNB, ALNC or both, for invasive BC between January 2014 and January 2024. A total of 179 patients were selected to ensure a representative cohort encompassing the three BC molecular subtypes (i.e. ER positive, HER2 positive and TNBC), as well as varying axillary (nodal) involvement and responses to NACT. For each case, the following clinico-pathological data were collected from the electronic patient record: age, molecular subtype, pathological TNM stage, histological grade, NACT received, response to NACT, lymph node procedure performed, size of primary tumor, size of largest ALN metastasis and overall survival ( Table 1 ). In order to calculate tumor burden in the axilla, we adapted the formula used to calculate residual cancer burden proposed by Symmans et al. 30 : tumor volume in the axilla = (1 − 0.75 LN ) d met ) , where LN is the number of positive lymph nodes and d met is the diameter of the largest nodal metastasis. Overall survival was defined as time from diagnosis to death of any cause. An additional 23 patients without a diagnosis of breast or any other cancer were also identified. These patients had undergone a lymph node biopsy or excision for non-malignant indications and were ultimately diagnosed with benign conditions, including follicular hyperplasia or reactive lymphadenopathy. These lymph nodes served as benign, reactive controls. Diagnostic slides were reviewed by Histopathologists (AL and KN) to confirm ALN disease status and select appropriate blocks for immunohistochemical and digital analysis. In each case, where available, a lymph node containing a metastastic deposit and a negative node were chosen. In cases with both micrometastasis and macrometastasis, examples of both were selected. In patients who had SLNB followed by ALNC, nodes from both procedures were included. Negative nodes from patients who underwent ALNC were not ordered or labeled by nodal level during surgery, and therefore were considered representative of the entire axillary basin. Each tissue sample, together with the associated data, was de-identified before being released to researchers to ensure blinded downstream analysis. METHOD DETAILS Tinctorial Staining Sections were cut at 3 µm thickness, then deparaffinised in xylene and rehydrated through graded alcohols and distilled water. Reticulin fibers were visualised using a modified silver impregnation method. Slides were first oxidized in acidified potassium permanganate (equal parts 1% potassium permanganate and 0.3% sulfuric acid) for 7 min. After rinsing, sections were bleached in 1.5% oxalic acid until colorless (1 min), followed by immersion in 2.5% iron alum (ferric ammonium sulfate, 2.5g/100 mL distilled water) for 10 min. Fresh silver solution was prepared by combining 7mL 10% silver nitrate with 3 mL industrial methylated spirits (IMS), followed by dropwise addition of 0.88 ammonia until the initial precipitate dissolved, then adding 5 additional drops of ammonia. The silver solution was applied to each section for 1 min. Slides were then developed for 1 min using a formaldehyde-based reticulin developer and stabilized in 5% sodium thiosulfate (2 min). Between each staining step, slides were rinsed with distilled water. Masson’s trichrome staining was performed using a celestine blue–haematoxylin sequence for nuclear visualization (5 min each), followed by tap water rinsing. Sections were then counterstained using a Ponceau–acid fuchsin solution for 5 min (prepared by mixing equal parts of Solution A and Solution B: Solution A, 0.5g Ponceau de xylidine in 1 mL glacial acetic acid and 100mL distilled water; Solution B, 0.5g acid fuchsin in 1mL glacial acetic acid and 100mL distilled water). After that, sections were differentiated in 1% phosphomolybdic acid for 4 min until collagen appeared colorless and counterstained with 1% light green SF for 2 min. Between each staining step, slides were rinsed with distilled water. Staining quality was microscopically confirmed at the differentiation and counterstaining stages. Following all staining protocols, sections were dehydrated, cleared in xylene, mounted and cover slipped. Immunohistochemical Staining Sections cut at 3 µm thickness were immunostained using either the Leica Bond III Autostainer (Leica Biosystems) or the Ventana Benchmark Ultra platform (Roche). Automated staining protocols were optimized to select the appropriate antigen retrieval method, antibody dilution and incubation period, and detection steps for each FRC antibody. Appropriate negative and positive control tissues were included with each batch. PDGFRβ was stained using a recombinant rabbit monoclonal antibody (clone RM303) at 1:100 dilution, with heat-induced epitope retrieval (HIER) in pH 9 buffer for 30 min on the Leica Bond III platform. Integrin β1 was detected using a rabbit monoclonal antibody (clone EP1041Y) at 1:300 dilution. Enzymatic antigen retrieval was performed using proteinase K for 10 min on the Leica Bond III platform. FAP (clone EPR20021) staining was performed at 1:100 dilution. HIER in pH 9 buffer for 30 min on Leica Bond III platform was the most effective, but results were not consistent. Podoplanin (clone D240) and SMA (clone 1A4) staining were performed using mouse monoclonal, ready to use, prediluted kits with HIER on the Ventana Benchmark Ultra platform. Region of Interest (ROI) Selection Whole-slide images of all stained sections were acquired using the Glissando Desktop Scanner (Objective Imaging). Digital images were viewed using QuPath (0.5.1-x64). For each PDGFRβ-stained lymph node section, ROIs measuring 250,000 µm² were selected at x20 magnification. In uninvolved (negative) nodes, eight ROI per section were manually selected from T cell zones, including a representative mix of paracortical and interfollicular areas, while avoiding germinal centers. In metastatic (positive) lymph nodes, eight ROI were selected from areas of metastasis. Nodes containing only isolated tumor cells, micrometastases or small macrometastases were excluded, as the deposits were smaller than the standardised ROI area. For each positive node, an additional eight ROI were selected from residual lymphoid tissue, at least 500 µm away from metastatic tumor deposits. Nodes that were completely replaced by tumor were excluded from this residual tissue analysis. In cases with histological evidence of NACT-induced fibrosis, four additional ROI were selected from fibrotic areas. Where fewer than the target number of ROI could be identified due to tissue limitations, the maximum feasible number was selected and documented. Image Analysis All ROI were analyzed using the TWOMBLI plugin for ImageJ (NIH) 27 to quantify spatial metrics of the FRC network. Prior to analysis, the blue haematoxylin counterstain was digitally removed from PDGFRβ-stained ROI images to minimize background noise. The following optimized parameters were applied to all images: contrast saturation = 0.35, minimum line width = 5, maximum line width = 20, minimum branch length = 10 and maximum display HDMI = 237. TWOMBLI derived outputs were: lacunarity (a measure of how completely the FRC network fills the ROI), HDMI (proportion of the ROI covered by FRC network), HGU (the number of endpoints per unit length), fiber alignment, FRC fiber length, FRC fiber width, the number of FRC endpoints and the number of FRC branchpoints. The same ROI images from metastatic ALN and areas of NACT-induced fibrosis were additionally analyzed using CurveAlign (v5.0, MATLAB-based), which is designed to accurately quantify fiber orientation and anisotropy. Cell counts within each ROI were performed using QuPath, with automated detection parameters optimized for each staining condition. QUANTIFICATION AND STATISTICAL ANALYSIS For each lymph node, the median value across the eight ROI was calculated for all quantitative parameters to minimize the impact of outliers. Data analysis was performed in RStudio (version 2024.09.1+394), using the following R packages: “tidyverse” for data manipulation and visualization, “ggplot2” for plotting, stats for regression analysis, and “survival” and “survminer” for survival analysis. Associations between TWOMBLI derived matrix features and clinical variables were assessed using multivariate linear regression. For comparisons between unmatched, non-parametric categorical subgroups, the Kruskal–Wallis test was performed in GraphPad Prism (version 10.4.1). A p-value ≤ 0.05 was considered statistically significant. The PCA included both TWOMBLI (i.e. lacunarity; branchpoints (normalized to fiber length); endpoints (normalized to fiber length); HDMI; and alignment) and clinical (i.e. NACT exposure; size/type of metastasis; pathological tumor stage; age; axillary tumor burden; and grade) parameters. Since some TWOMBLI parameters (i.e. HGU; fiber length; and width) were dependent on those listed above, they were excluded from the PCA. Survival analysis was conducted in RStudio using the Kaplan–Meier method; group comparisons were performed using the Gehan-Breslow-Wilcoxon method. ACKNOWLEDGMENTS We would like to thank the patients for consenting to the use of their tissue for this project. This work was funded by Breast Cancer Now (Dame Vera Lynn Breast Cancer Now Clinical Research Training Fellowship 2023.06CRTF1643, A.L) and Cancer Research UK (RCCSCF-May22\100001, S.E.A). Funder Information Declared Breast Cancer Now , 2023.06CRTF1643 Cancer Research UK , RCCSCF-May22\100001 REFERENCES 1. ↵ DiSipio , T. , Rye , S. , Newman , B. , and Hayes , S . ( 2013 ). Incidence of unilateral arm lymphoedema after breast cancer: a systematic review and meta-analysis . Lancet Oncol 14 , 500 – 515 . doi: 10.1016/S1470-2045(13)70076-7 . OpenUrl CrossRef PubMed 2. ↵ Henke , G. , Knauer , M. , Ribi , K. , Hayoz , S. , Gérard , M.-A. , Ruhstaller , T. , Zwahlen , D.R. , Muenst , S. , Ackerknecht , M. , Hawle , H. , et al. ( 2018 ). Tailored axillary surgery with or without axillary lymph node dissection followed by radiotherapy in patients with clinically node-positive breast cancer (TAXIS): study protocol for a multicenter, randomized phase-III trial . Trials 19 , 667 . doi: 10.1186/s13063-018-3021-9 . OpenUrl CrossRef 3. ↵ Giuliano , A.E. , Ballman , K.V. , McCall , L. , Beitsch , P.D. , Brennan , M.B. , Kelemen , P.R. , Ollila , D.W. , Hansen , N.M. , Whitworth , P.W. , Blumencranz , P.W. , et al. ( 2017 ). Effect of Axillary Dissection vs No Axillary Dissection on 10-Year Overall Survival Among Women With Invasive Breast Cancer and Sentinel Node Metastasis: The ACOSOG Z0011 (Alliance) Randomized Clinical Trial . JAMA 318 , 918 . doi: 10.1001/jama.2017.11470 . OpenUrl CrossRef PubMed 4. ↵ Bartels , S.A.L. , Donker , M. , Poncet , C. , Sauvé , N. , Straver , M.E. , van de Velde , C.J.H. , Mansel , R.E. , Blanken , C. , Orzalesi , L. , Klinkenbijl , J.H.G. , et al. ( 2023 ). Radiotherapy or Surgery of the Axilla After a Positive Sentinel Node in Breast Cancer: 10-Year Results of the Randomized Controlled EORTC 10981-22023 AMAROS Trial . Journal of Clinical Oncology 41 , 2159 – 2165 . doi: 10.1200/JCO.22.01565 . OpenUrl CrossRef PubMed 5. ↵ de Boer , M. , van Deurzen , C.H.M. , van Dijck , J.A.A.M. , Borm , G.F. , van Diest , P.J. , Adang , E.M.M. , Nortier , J.W.R. , Rutgers , E.J.T. , Seynaeve , C. , Menke-Pluymers , M.B.E. , et al. ( 2009 ). Micrometastases or isolated tumor cells and the outcome of breast cancer . N Engl J Med 361 , 653 – 663 . doi: 10.1056/NEJMoa0904832 . OpenUrl CrossRef PubMed Web of Science 6. ↵ NICE ( 2022 ). Pembrolizumab for neoadjuvant and adjuvant treatment of triple negative early or locally advanced breast cancer https://www.nice.org.uk/guidance/ta851 . 7. ↵ Acton , S.E. , Onder , L. , Novkovic , M. , Martinez , V.G. , and Ludewig , B . ( 2021 ). Communication, construction, and fluid control: lymphoid organ fibroblastic reticular cell and conduit networks . Trends in Immunology 42 , 782 – 794 . doi: 10.1016/j.it.2021.07.003 . OpenUrl CrossRef PubMed 8. ↵ Anderson , A.O. , and Anderson , N.D . ( 1975 ). Studies on the structure and permeability of the microvasculature in normal rat lymph nodes . American Journal of Pathology 80 , 387 – 418 . OpenUrl PubMed Web of Science 9. Gretz , J.E. , Norbury , C.C. , Anderson , A.O. , Proudfoot , A.E. , and Shaw , S . ( 2000 ). Lymph-borne chemokines and other low molecular weight molecules reach high endothelial venules via specialized conduits while a functional barrier limits access to the lymphocyte microenvironments in lymph node cortex . J Exp Med 192 , 1425 – 1440 . doi: 10.1084/jem.192.10.1425 . OpenUrl Abstract / FREE Full Text 10. ↵ Sixt , M. , Kanazawa , N. , Selg , M. , Samson , T. , Roos , G. , Reinhardt , D.P. , Pabst , R. , Lutz , M.B. , and Sorokin , L . ( 2005 ). The conduit system transports soluble antigens from the afferent lymph to resident dendritic cells in the T cell area of the lymph node . Immunity 22 , 19 – 29 . doi: 10.1016/j.immuni.2004.11.013 . OpenUrl CrossRef PubMed Web of Science 11. ↵ Rodda , L.B. , Lu , E. , Bennett , M.L. , Sokol , C.L. , Wang , X. , Luther , S.A. , Barres , B.A. , Luster , A.D. , Ye , C.J. , and Cyster , J.G . ( 2018 ). Single-Cell RNA Sequencing of Lymph Node Stromal Cells Reveals Niche-Associated Heterogeneity . Immunity 48 , 1014 – 1028 .e1016. doi: 10.1016/j.immuni.2018.04.006 . OpenUrl CrossRef PubMed 12. ↵ Acton , S.E. , Astarita , J.L. , Malhotra , D. , Lukacs-Kornek , V. , Franz , B. , Hess , P.R. , Jakus , Z. , Kuligowski , M. , Fletcher , A.L. , Elpek , K.G. , et al. ( 2012 ). Podoplanin-rich stromal networks induce dendritic cell motility via activation of the C-type lectin receptor CLEC-2 . Immunity 37 , 276 – 289 . doi: 10.1016/j.immuni.2012.05.022 . OpenUrl CrossRef PubMed 13. Astarita , J.L. , Cremasco , V. , Fu , J. , Darnell , M.C. , Peck , J.R. , Nieves-Bonilla , J.M. , Song , K. , Kondo , Y. , Woodruff , M.C. , Gogineni , A. , et al. ( 2015 ). The CLEC-2-podoplanin axis controls the contractility of fibroblastic reticular cells and lymph node microarchitecture . Nat Immunol 16 , 75 – 84 . doi: 10.1038/ni.3035 . OpenUrl CrossRef PubMed 14. ↵ Makris , S. , Hari-Gupta , Y. , Cantoral-Rebordinos , J.A. , Martinez , V.G. , Horsnell , H.L. , Benjamin , A.C. , Cinti , I. , Jovancheva , M. , Shewring , D. , Nguyen , N. , et al. ( 2024 ). Lymph Node Fibroblast Phenotypes and Immune Crosstalk Regulated by Podoplanin Activity . bioRxiv . 15. ↵ Naidoo , K. , and Pinder , S.E . ( 2017 ). Micro- and macro-metastasis in the axillary lymph node: A review . Surgeon 15 , 76 – 82 . doi: 10.1016/j.surge.2016.07.002 . OpenUrl CrossRef PubMed 16. ↵ Fletcher , A.L. , Malhotra , D. , Acton , S.E. , Lukacs-Kornek , V. , Bellemare-Pelletier , A. , Curry , M. , Armant , M. , and Turley , S.J . ( 2011 ). Reproducible Isolation of Lymph Node Stromal Cells Reveals Site-Dependent Differences in Fibroblastic Reticular Cells . Frontiers in Immunology 2 . doi: 10.3389/fimmu.2011.00035 . OpenUrl CrossRef PubMed 17. ↵ Roet , J.E.G. , Morrison , A.I. , Mikula , A.M. , de Kok , M. , Panocha , D. , Roest , H.P. , van der Laan , L.J.W. , de Winde , C.M. , and Mebius , R.E. ( 2024 ). Human lymph node fibroblastic reticular cells maintain heterogeneous characteristics in culture . iScience 27 , 110179 . doi: 10.1016/j.isci.2024.110179 . OpenUrl CrossRef PubMed 18. ↵ Sahai , E. , Astsaturov , I. , Cukierman , E. , DeNardo , D.G. , Egeblad , M. , Evans , R.M. , Fearon , D. , Greten , F.R. , Hingorani , S.R. , Hunter , T. , et al. ( 2020 ). A framework for advancing our understanding of cancer-associated fibroblasts . Nat Rev Cancer 20 , 174 – 186 . doi: 10.1038/s41568-019-0238-1 . OpenUrl CrossRef PubMed 19. ↵ Severino , P. , Palomino , D.T. , Alvarenga , H. , Almeida , C.B. , Pasqualim , D.C. , Cury , A. , Salvalaggio , P.R. , De Vasconcelos Macedo , A.L. , Andrade , M.C. , Aloia , T. , et al. ( 2017 ). Human Lymph Node-Derived Fibroblastic and Double-Negative Reticular Cells Alter Their Chemokines and Cytokines Expression Profile Following Inflammatory Stimuli . Frontiers in Immunology 8 . doi: 10.3389/fimmu.2017.00141 . OpenUrl CrossRef 20. Valencia , J. , Jiménez , E. , Martínez , V.G. , Del Amo , B.G. , Hidalgo , L. , Entrena , A. , Fernández-Sevilla , L.M. , Del Río , F. , Varas , A. , Vicente , Á. , and Sacedón , R . ( 2017 ). Characterization of human fibroblastic reticular cells as potential immunotherapeutic tools . Cytotherapy 19 , 640 – 653 . doi: 10.1016/j.jcyt.2017.01.010 . OpenUrl CrossRef PubMed 21. ↵ Knoblich , K. , Cruz Migoni , S. , Siew , S.M. , Jinks , E. , Kaul , B. , Jeffery , H.C. , Baker , A.T. , Suliman , M. , Vrzalikova , K. , Mehenna , H. , et al. ( 2018 ). The human lymph node microenvironment unilaterally regulates T-cell activation and differentiation . PLoS Biol 16 , e2005046 . doi: 10.1371/journal.pbio.2005046 . OpenUrl CrossRef PubMed 22. ↵ Mechthild Lütge , L.K. , Yves Stanossek , Samuel Meili , Hung-Wei Cheng , Angelina De Martin , Joshua Brandstadter , Ivan Maillard , Mark D. Robinson , Sandro J. Stoeckli , Natalia B. Pikor , Lucas Onder , Burkhard Ludewig ( 2025 ). Fibroblastic reticular cells form reactive myeloid cell niches in human lymph nodes . Sci. Immunol . 10 . doi: 10.1126/sciimmunol.ads6820 . OpenUrl CrossRef 23. ↵ Mackenzie , D.H . ( 1959 ). Reticulin Patterns in Tumours of Lymphoid Tissue . Br J Cancer 13 , 38 – 44 . OpenUrl PubMed 24. ↵ Bancroft , J.D. , and Gamble , M. ( 2008 ). Theory and Practice of Histological Techniques , Edinburgh Edition ( Churchill Livingstone/Elsevier ). 25. ↵ Pelon , F. , Bourachot , B. , Kieffer , Y. , Magagna , I. , Mermet-Meillon , F. , Bonnet , I. , Costa , A. , Givel , A.-M. , Attieh , Y. , Barbazan , J. , et al. ( 2020 ). Cancer-associated fibroblast heterogeneity in axillary lymph nodes drives metastases in breast cancer through complementary mechanisms . Nat Commun 11 , 404 . doi: 10.1038/s41467-019-14134-w . OpenUrl CrossRef PubMed 26. ↵ Costa , A. , Kieffer , Y. , Scholer-Dahirel , A. , Pelon , F. , Bourachot , B. , Cardon , M. , Sirven , P. , Magagna , I. , Fuhrmann , L. , Bernard , C. , et al. ( 2018 ). Fibroblast Heterogeneity and Immunosuppressive Environment in Human Breast Cancer . Cancer Cell 33 , 463 – 479 .e410. doi: 10.1016/j.ccell.2018.01.011 . OpenUrl CrossRef PubMed 27. ↵ Wershof , E. , Park , D. , Barry , D.J. , Jenkins , R.P. , Rullan , A. , Wilkins , A. , Schlegelmilch , K. , Roxanis , I. , Anderson , K.I. , Bates , P.A. , and Sahai , E . ( 2021 ). A FIJI macro for quantifying pattern in extracellular matrix . Life Sci Alliance 4 , e202000880 . doi: 10.26508/lsa.202000880 . OpenUrl Abstract / FREE Full Text 28. ↵ Pinder , S.E. , Rakha , E.A. , Purdie , C.A. , Bartlett , J.M.S. , Francis , A. , Stein , R.C. , Thompson , A.M. , Shaaban , A.M. , and Group , t.T.S.o.t.N.B.C.S. ( 2015 ). Macroscopic handling and reporting of breast cancer specimens pre- and post-neoadjuvant chemotherapy treatment: review of pathological issues and suggested approaches . Histopathology 67 , 279 – 293 . doi: 10.1111/his.12649 . OpenUrl CrossRef PubMed 29. ↵ Liu , Y. , Keikhosravi , A. , Mehta , G.S. , Drifka , C.R. , and Eliceiri , K.W . ( 2017 ). Methods for Quantifying Fibrillar Collagen Alignment . Methods Mol Biol 1627 , 429 – 451 . doi: 10.1007/978-1-4939-7113-8_28 . OpenUrl CrossRef 30. ↵ Symmans , W.F. , Peintinger , F. , Hatzis , C. , Rajan , R. , Kuerer , H. , Valero , V. , Assad , L. , Poniecka , A. , Hennessy , B. , Green , M. , et al. ( 2007 ). Measurement of residual breast cancer burden to predict survival after neoadjuvant chemotherapy . J Clin Oncol 25 , 4414 – 4422 . doi: 10.1200/JCO.2007.10.6823 . OpenUrl Abstract / FREE Full Text 31. ↵ Onder , L. , Papadopoulou , C. , Lutge , A. , Cheng , H.W. , Lutge , M. , Perez-Shibayama , C. , Gil-Cruz , C. , De Martin , A. , Kurz , L. , Cadosch , N. , et al. ( 2025 ). Fibroblastic reticular cells generate protective intratumoral T cell environments in lung cancer . Cell 188 , 430 – 446 e420. doi: 10.1016/j.cell.2024.10.042 . OpenUrl CrossRef 32. ↵ Martinez , V.G. , Pankova , V. , Krasny , L. , Singh , T. , Makris , S. , White , I.J. , Benjamin , A.C. , Dertschnig , S. , Horsnell , H.L. , Kriston-Vizi , J. , et al. ( 2019 ). Fibroblastic Reticular Cells Control Conduit Matrix Deposition during Lymph Node Expansion . Cell Reports 29 , 2810 – 2822 .e2815. doi: 10.1016/j.celrep.2019.10.103 . OpenUrl CrossRef PubMed 33. ↵ Fidler , I . ( 2003 ). The pathogenesis of cancer metastasis: the ‘seed and soil’ hypothesis revisited . Nat Rev Cancer 3 , 453 – 458 . doi: 10.1038/nrc1098 . OpenUrl CrossRef PubMed Web of Science 34. ↵ NICE ( 2025 ). Early and locally advanced breast cancer: diagnosis and management. https://www.nice.org.uk/guidance/ng101 . 35. ↵ Riedel , A. , Shorthouse , D. , Haas , L. , Hall , B.A. , and Shields , J . ( 2016 ). Tumor-induced stromal reprogramming drives lymph node transformation . Nat Immunol 17 , 1118 – 1127 . doi: 10.1038/ni.3492 . OpenUrl CrossRef PubMed 36. ↵ Salgado , R. , Denkert , C. , Demaria , S. , Sirtaine , N. , Klauschen , F. , Pruneri , G. , Wienert , S. , Van den Eynden , G. , Baehner , F.L. , Penault-Llorca , F. , et al. ( 2015 ). The evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer: recommendations by an International TILs Working Group 2014 . Ann Oncol 26 , 259 – 271 . doi: 10.1093/annonc/mdu450 . OpenUrl CrossRef PubMed 37. ↵ Grigoriadis , A. , Gazinska , P. , Pai , T. , Irhsad , S. , Wu , Y. , Millis , R. , Naidoo , K. , Owen , J. , Gillett , C.E. , Tutt , A. , et al. ( 2018 ). Histological scoring of immune and stromal features in breast and axillary lymph nodes is prognostic for distant metastasis in lymph node-positive breast cancers: Prognostic value of histological immune and stromal features . J Path: Clin Res 4 , 39 – 54 . doi: 10.1002/cjp2.87 . OpenUrl CrossRef 38. ↵ Voorwerk , L. , Slagter , M. , Horlings , H.M. , Sikorska , K. , van de Vijver , K.K. , de Maaker , M. , Nederlof , I. , Kluin , R.J.C. , Warren , S. , Ong , S. , et al. ( 2019 ). Immune induction strategies in metastatic triple-negative breast cancer to enhance the sensitivity to PD-1 blockade: the TONIC trial . Nat Med 25 , 920 – 928 . doi: 10.1038/s41591-019-0432-4 . OpenUrl CrossRef PubMed 39. ↵ Schmid , P. , Cortes , J. , Pusztai , L. , McArthur , H. , Kümmel , S. , Bergh , J. , Denkert , C. , Park , Y.H. , Hui , R. , Harbeck , N. , et al. ( 2020 ). Pembrolizumab for Early Triple-Negative Breast Cancer . New England Journal of Medicine 382 , 810 – 821 . doi: 10.1056/NEJMoa1910549 . OpenUrl CrossRef PubMed 40. ↵ Boyle , M.K. , Amersi , F. , Chung , A. , Tseng , J. , and Giuliano , A.E . ( 2025 ). Comparison of Targeted Axillary Dissection with Sentinel Node Biopsy Alone on Nodal Recurrence for Patients who have Node-Positive Breast Cancer Treated with Neoadjuvant Chemotherapy . Ann Surg Oncol . doi: 10.1245/s10434-025-17197-w . OpenUrl CrossRef 41. ↵ Novkovic , M. , Onder , L. , Cupovic , J. , Abe , J. , Bomze , D. , Cremasco , V. , Scandella , E. , Stein , J.V. , Bocharov , G. , Turley , S.J. , and Ludewig , B . ( 2016 ). Topological Small-World Organization of the Fibroblastic Reticular Cell Network Determines Lymph Node Functionality . PLoS Biol 14 , e1002515 . doi: 10.1371/journal.pbio.1002515 . OpenUrl CrossRef PubMed 42. ↵ Francis , K. , and Palsson , B . ( 1997 ). Effective intercellular communication distances are determined by the relative time constants for cyto chemokine secretion and diffusion . Proc Natl Acad Sci U S A 94 . doi: 10.1073/pnas.94.23.12258 . OpenUrl Abstract / FREE Full Text 43. ↵ Swartz , M.A. , and Lund , A.W . ( 2012 ). Lymphatic and interstitial flow in the tumour microenvironment: linking mechanobiology with immunity . Nat Rev Cancer 12 , 210 – 219 . doi: 10.1038/nrc3186 . OpenUrl CrossRef PubMed Web of Science 44. ↵ Stevenson , J. , Barrow-McGee , R. , Yu , L. , Paul , A. , Mansfield , D. , Owen , J. , Woodman , N. , Natrajan , R. , Haider , S. , Gillett , C. , et al. ( 2021 ). Proteomics of REPLICANT perfusate detects changes in the metastatic lymph node microenvironment . npj Breast Cancer 7 , 24 . doi: 10.1038/s41523-021-00227-7 . OpenUrl CrossRef 45. Barrow-McGee , R. , Procter , J. , Owen , J. , Woodman , N. , Lombardelli , C. , Kothari , A. , Kovacs , T. , Douek , M. , George , S. , Barry , P.A. , et al. ( 2020 ). Real-time ex vivo perfusion of human lymph nodes invaded by cancer (REPLICANT): a feasibility study . The Journal of Pathology 250 , 262 – 274 . doi: 10.1002/path.5367 . OpenUrl CrossRef PubMed 46. ↵ Llewellyn , A. , Barrow-McGee , R. , Stevenson , J. , Gore , J. , and Naidoo , K . ( 2025 ). Protocol for perfusing human axillary lymph nodes ex vivo to study structure and function in real time . STAR Protocols 6 , 103624 . doi: 10.1016/j.xpro.2025.103624 . OpenUrl CrossRef 47. Bankhead , P. , Loughrey , M.B. , Fernandez , J.A. , Dombrowski , Y. , McArt , D.G. , Dunne , P.D. , McQuaid , S. , Gray , R.T. , Murray , L.J. , Coleman , H.G. , et al. ( 2017 ). QuPath: Open source software for digital pathology image analysis . Sci Rep 7 , 16878 . doi: 10.1038/s41598-017-17204-5 . OpenUrl CrossRef PubMed 48. Schneider , C.A. , Rasband , W.S. , and Eliceiri , K.W . ( 2012 ). NIH Image to ImageJ: 25 years of image analysis . Nat Methods 9 , 671 – 675 . doi: 10.1038/nmeth.2089 . OpenUrl CrossRef PubMed Web of Science View the discussion thread. Back to top Previous Next Posted July 09, 2025. Download PDF Supplementary Material Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Topological Analysis of the Human Lymph Node Reticular Network Predicts Outcome in Breast Cancer Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Topological Analysis of the Human Lymph Node Reticular Network Predicts Outcome in Breast Cancer Amy Llewellyn , Sharon D’Costa , Ringo Lam , Jasmine Gore , Veronika Lachina , Daniel Shewring , Sophie E. Acton , Kalnisha Naidoo bioRxiv 2025.07.07.663481; doi: https://doi.org/10.1101/2025.07.07.663481 Share This Article: Copy Citation Tools Topological Analysis of the Human Lymph Node Reticular Network Predicts Outcome in Breast Cancer Amy Llewellyn , Sharon D’Costa , Ringo Lam , Jasmine Gore , Veronika Lachina , Daniel Shewring , Sophie E. Acton , Kalnisha Naidoo bioRxiv 2025.07.07.663481; doi: https://doi.org/10.1101/2025.07.07.663481 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Pathology Subject Areas All Articles Animal Behavior and Cognition (7622) Biochemistry (17648) Bioengineering (13871) Bioinformatics (41880) Biophysics (21423) Cancer Biology (18561) Cell Biology (25461) Clinical Trials (138) Developmental Biology (13364) Ecology (19866) Epidemiology (2067) Evolutionary Biology (24290) Genetics (15590) Genomics (22475) Immunology (17713) Microbiology (40328) Molecular Biology (17148) Neuroscience (88473) Paleontology (666) Pathology (2827) Pharmacology and Toxicology (4816) Physiology (7635) Plant Biology (15114) Scientific Communication and Education (2044) Synthetic Biology (4286) Systems Biology (9815) Zoology (2268)