Intravenous leiomyomatosis: A retrospective case series of imaging findings and diagnostic challenges.

A 37-year-old woman with a prior history of myomectomy and cesarean section presented with a constellation of symptoms, including back pain, menorrhagia, and bilateral lower extremity edema. Preoperative imaging (ultrasound and contrast-enhanced CT/MRI) revealed a markedly enlarged uterus with heterogeneous myometrial thickening and indistinct boundaries between the uterus and the bladder wall, along with multiple pelvic and retroperitoneal solid masses. Tumor involvement was noted in bilateral ovarian veins and right internal iliac vein → common iliac vein → IVC → right atrium, concurrent with bilateral giant multilocular ovarian cystic masses. Initial diagnosis suggested uterine adenomyosis, IVL, and bilateral ovarian mucinous cystadenomas ( Fig. 4 ). Subsequent PET-CT demonstrated markedly elevated 18F-FDG avidity, with a maximum standardized uptake value (SUVmax) of 20.78, raising strong suspicion for malignancy. Final histopathology confirmed low-grade ESS with full-thickness myometrial infiltration, bilateral adnexal involvement, and widespread metastatic extension to retroperitoneal, pelvic, and intravascular sites. The patient’s serum CA125 level (106.5 U/mL) was significantly higher than that of the 13 IVL cases (15.77 U/mL, 95% CI: 10.72-28.71). Fig. 4 ESS case. (A) Contrast-enhanced sagittal CT imaging demonstrates multiple cystic/solid masses (arrows) of varying sizes located in the pelvic cavity and retroperitoneum. (B and C) PET-CT demonstrated markedly elevated 18F-FDG avidity in these lesions, with the bilateral ovarian veins were involved (C, arrows). (D) Histopathology (H&E) revealed tumor cells with short spindle morphology and oval nuclei, exhibiting minimal cytological atypia and visible mitotic figures (2-3/10 HPF) (Immunoprofile: CD10 (diffuse)+, FH/Vim/ER/PR+, Caldesmon/CyclinD1/P16 (focal/weak)+, Ki67∼10%, P53(wild-type), CD31/ERG (vascular +), Desmin/Neg/ALK (D5F3)/BCOR/TRK/Myogenin−). Fig 4 dummy alt text ESS case. (A) Contrast-enhanced sagittal CT imaging demonstrates multiple cystic/solid masses (arrows) of varying sizes located in the pelvic cavity and retroperitoneum. (B and C) PET-CT demonstrated markedly elevated 18F-FDG avidity in these lesions, with the bilateral ovarian veins were involved (C, arrows). (D) Histopathology (H&E) revealed tumor cells with short spindle morphology and oval nuclei, exhibiting minimal cytological atypia and visible mitotic figures (2-3/10 HPF) (Immunoprofile: CD10 (diffuse)+, FH/Vim/ER/PR+, Caldesmon/CyclinD1/P16 (focal/weak)+, Ki67∼10%, P53(wild-type), CD31/ERG (vascular +), Desmin/Neg/ALK (D5F3)/BCOR/TRK/Myogenin−).

Bing Wang and Guanxun Cheng: Conceptualization, Methodology. Bing Wang and Yimei Liao: Data curation, Writing − Original draft preparation. Yulin Li, JunSheng Jiang and Haibo Chen: Visualization and Investigation. Jiaji Mao, Baoxun Li and Zhidong Yuan:Investigation and Resources. Renzhang Tang:Application of statistical analysis. Ying Yuan, Congying Chen and Guanxun Cheng: Supervision and Validation, interpretation of data for the work. Minghui Cao and Yimei Liao: Writing − Reviewing and Editing. All authors read and approved the final manuscript.

This retrospective observational study was conducted in accordance with the Declaration of Helsinki and was approval by the Ethics Committee of Peking University Shenzhen Hospital (IRB No. 2025 [125]), Shenzhen Maternity and Child Healthcare Hospital (IRB No. SFYLS [2025]056) and Sun Yat-Sen Memorial Hospital (IRB No. SYSKY-2024-146-01).

Imaging characteristics were systematically analyzed and categorized by anatomical distribution (pelvic vs intravascular lesions). For pelvic lesions: solid lesions were defined as those with no internal fluid areas; predominantly solid lesions as those containing >50% solid tissue; and predominantly cystic lesions as those with >50% cystic content, some of which showed only focal thickening of the cyst wall. For intravascular lesions: solid lesions appeared as irregular, elongated solid densities (or echogenicities) without internal fluid components; sieve-like intravascular lesions demonstrated numerous internal voids on axial images, while coronal views revealed a characteristic “loofah sponge” morphology. Disease staging followed the Ma et al. [ 4 ] vascular progression criteria: Stage I, lesions confined to the uterus and parauterine region, limited to the pelvis without abdominal major vessel involvement; Stage II, lesions invading abdominal vessels, entering the IVC but not reaching the renal vein level; Stage III, tumors extending beyond the renal vein level, potentially reaching the right atrium but not the pulmonary artery; Stage IV, tumors extending into the pulmonary artery and/or developing lung metastases. Vascular dissemination patterns were further classified as originating from the ovarian veins and/or uterine–iliac venous pathways [ 4 ].

Written informed consent was obtained from the patient for the publication of the present case report and any accompanying images and related investigations.

Our study highlights the substantial diagnostic complexity of IVL, particularly given the high proportion of asymptomatic cases (47.3%) and nonspecific clinical presentations. These challenges often delay diagnosis and underscore the critical role of imaging in establishing a definitive diagnosis and guiding surgical strategy. We recommend a 2-pronged imaging approach: first, characterizing the primary pelvic lesion, and second, systematically assessing intravascular tumor extension. This approach reflects IVL’s typical pathogenesis, characterized by pelvic origin with subsequent venous system extension. For pelvic lesions, we propose categorizing IVL into 3 types—solid, predominantly solid, and predominantly cystic. The masses may arise within the myometrium, bulge outward toward the serosal surface, extend into the uterine cavity, or occupy the paracervical space, broad ligament, or adnexal region [ 7 ]. In our series, solid and predominantly solid masses were the most common sonographic patterns, typically appearing as irregular, beaded, nodular, or cord-like masses along the uterine contour. The solid-predominant lesions often displayed internal or peripheral anechoic regions (irregular or crescent-shaped), a characteristic feature of IVL. On CT, these corresponded to irregular fluid-density areas; on MRI, heterogeneous mixed signals revealed worm-like structures within the mass, consistent with intravascular tumor extension [ 10 ]. A subset of lesions were predominantly cystic (10.5%), representing an atypical manifestation that may be related to cystic degeneration or concurrent endometriosis. These lesions frequently contained solid components or thick internal septations, with color Doppler showing short linear or dotted blood flow signals within solid components or the cyst wall. Corresponding contrast-enhanced CT or MRI typically revealed heterogeneous enhancement of these areas. Due to these features, predominantly cystic IVL lesions are at high risk of being misdiagnosed as cystic neoplasms, encapsulated effusion, chocolate cysts, or tubal effusion. For intravascular lesions, a key diagnostic clue is worm-like extension of a uterine mass into adjacent parametrial veins—considered a pathognomonic sign of early IVL. However, this classic sign was clearly observed in only 3 of 19 cases (15.8%) in our cohort, highlighting its limited sensitivity. Intravascular IVL lesions can present as either solid or sieve-like structures [ 11 , 12 ]. In our series, 77.8% were solid (irregular and elongated densities without internal fluid) and 22.2% demonstrated sieve-like morphology, which is consistent with the findings reported by Ma et al. [ 13 ]. The slow venous growth of IVL, along with preserved peripheral blood flow around solid tumors or through sieve-like architectures, may explain the rarity of circulatory obstruction in stages II-IV disease despite extensive vascular involvement. Previous literature describes 4 patterns of IVL vascular extension: (1) the most common route—parametrial uterine vein → internal iliac vein → common iliac vein → IVC → right atrium [ 4 , 8 ]; (2) extension solely via the ovarian vein, either through the left ovarian vein into the left renal vein and IVC, or directly from the right ovarian vein into the IVC; (3) simultaneous extension via both pathways—observed in 47.4% of our cohort, reinforcing the importance of evaluating both ovarian and internal iliac veins preoperatively; and (4) rare discontinuous growth within the venous system or cardiac [ 14 ], or metastatic dissemination, presenting as multiple pulmonary nodules. In our series, 1 case demonstrated IVL extending into the IVC without cardiac involvement but with bilateral lung metastases, confirming metastatic potential. Pulmonary artery involvement occurs in approximately 3.9% of cases [ 4 ], with distal branch lesions often appearing as multiple lower lobe nodules that may mimic parenchymal disease. For optimal preoperative assessment, we advocate for a multimodal strategy tailored to disease progression patterns [ 11 ]. Ultrasound remains ideal for initial screening and postoperative surveillance due to its cost-effectiveness, convenience, and lack of radiation [ 7 , 15 ]. However, its diagnostic accuracy varies widely (14.3%-61.1%) for IVL [ 7 , 8 , 16 ], particularly limited in large pelvic masses where transvaginal imaging cannot evaluate parametrial vasculature and transabdominal imaging may be compromised by bowel gas. Contrast-enhanced CT has a lower miss rate (14.9%) [ 16 ], and remains the primary modality for detecting venous-phase filling defects. MRI provides superior soft-tissue resolution: T2-weighted sequences excel at demonstrating pathognomonic worm-like intratumoral contours [ 10 , 11 ] and detecting subtle parametrial venous involvement [ 8 ]. Thus, CT and MRI are superior for definitive diagnosis and preoperative planning. IVL is frequently misdiagnosed—most commonly as venous thrombosis [ 6 ]—with reported misdiagnosis rates of 30.8% for both ultrasound and CT (up to 94.4% for myometrium-confined cases [ 7 ]), and missed diagnosis rates of 38.8% and 14.9%, respectively [ 16 ]. In our study, 57.9% of patients experienced misdiagnosed or missed preoperatively during their extended diagnostic odyssey. Key diagnostic pitfalls included: (1) atypical radiological presentations (e.g., multilocular cystic masses mimicking cystadenoma); (2) diagnostic confusion caused by coexisting pelvic AVM; (3) differentiation from versus thrombosis; (4) diagnostic overshadowing by concurrent intestinal perforation; (5) omission of ovarian vein involvement; and (6) large tumor size limiting ultrasound detection of parametrial vascular involvement. IVL with AVMs is rare but distinct; arterial-phase enhancement of the IVC due to AVM makes intraluminal filling defects more conspicuous [ 17 ], unlike conventional IVL which is venous-phase predominant. In such cases, preoperative AVM embolization may be crucial to minimize bleeding risk [ 17 , 18 ]. Differentiating IVL from ESS is crucial, as illustrated by a case in our cohort initially concerning for IVL but ultimately diagnosed as ESS due to overlapping imaging features. However, this case exhibited malignant features atypical for IVL, including retroperitoneal dissemination, bladder wall invasion, elevated CA125 (106.5 U/mL), and markedly increased 18F-FDG avidity on PET-CT (SUV max 20.78). In contrast, IVL typically demonstrates only mild metabolic activity (SUV max 1.6-2.1) [ 19 ], reflecting its benign nature. Based on the imaging features observed in our series and reported in the literature, we propose analyzing IVL lesions by their pelvic and vascular components. Furthermore, we recommend stratifying imaging findings into typical and atypical presentations, with a particular focus on 2 underrecognized atypical features: cystic-predominant pelvic masses and sieve-like intravascular lesions. This framework may facilitate earlier recognition of this rare disease. Given its rarity, future multicenter studies with larger patient cohorts are needed to validate this classification and establish a reproducible, clinically actionable consensus.

During the preparation of this work the authors did not used AI-tools.

Intravenous leiomyomatosis (IVL) is a rare smooth muscle neoplasm that is histologically benign but biologically aggressive, arising from uterine leiomyomas or the smooth muscle of uterine vein walls. Its distinguishing characteristic is intraluminal venous extension, in which tumor proliferate within uterine, ovarian or iliac veins and may extend proximally into the inferior vena cava (IVC), right heart and pulmonary arteries, producing a spectrum of clinical manifestations [ [1] , [2] , [3] , [4] ]. A systematic review reported cardiac involvement in approximately half of patients and an overall mortality of around 2.4% [ 5 ]. Because early-stage disease often mimics a typical uterine leiomyoma on imaging and symptoms are nonspecific, IVL is frequently misdiagnosed as benign fibroids, gynecologic malignancies, venous thrombosis or metastatic disease [ [6] , [7] , [8] ]. Consequently, despite its original description in 1896 [ 9 ], IVL remains diagnostically and therapeutically challenging, with most available evidence derived from case reports and small series. Complete surgical resection is considered the cornerstone of treatment [ 4 , 5 ], but its success critically depends on accurate preoperative identification of the full extent of intravascular tumor. Contrast-enhanced CT, MRI, and ultrasound are commonly used for IVL diagnosis; however, systematic assessment of their performance—particularly multimodal image analysis that simultaneously targets both the primary pelvic lesions and intravascular extensions—has yet to be fully elucidated. To address this gap, we conducted a decade-long retrospective study of IVL cases from 2 tertiary centers, analyzing clinical and multimodal imaging data (ultrasound, CT, and MRI). We aimed to characterize typical and atypical imaging features of pelvic and intravascular IVL components, identify causes of misdiagnosis, and develop imaging-guided strategies for complete tumor mapping and surgical planning.

All patients underwent gynecologic ultrasound, which revealed uterine or pelvic masses with a mean maximum diameter of 8.99 ± 5.65 cm (range: 1.5-25.0 cm). Only 3 cases (15.8%) demonstrated the characteristic worm-like extension of an intramural leiomyoma into the parametrial veins (Supplementary Information 1: Video 1 ), a typical and critical early diagnostic feature of IVL. Nine cases exhibited irregular, beaded, or cord-like masses in the parametrial or pelvic region ( Fig. 1 A, B, and E) with either well-defined or ill-defined demarcation relative to the uterine serosa. Among the 4 recurrent post-hysterectomy cases, all presented with multifocal pelvic lesions. Ultrasound failed to detect pelvic involvement in 3 of these 4 cases (15.8% of the overall cohort). These included 1 case with myometrium-confined disease (identified only on pathology) and 2 cases in which massive tumor size obscured parametrial involvement. Fig. 1 (A-D) Transvaginal ultrasound demonstrates characteristic IVL morphologies: beaded (A, arrow), cord-like (B), and mass-like (C-E) pelvic lesions. Color Doppler reveals intralesional and perilesional punctate/short linear vascular signals. Solid-predominant masses display irregular (C, arrows) or crescent-shaped cystic components (D, arrow) within the hypoechoic tumor tissue. (E-H) Multimodal imaging in a case with posthysterectomy IVL recurrence. While ultrasound (E) and CT (F) depict heterogeneous pelvic masses, MRI T2WI (G and H) clearly delineates pathognomonic worm-like morphology (arrows). (I-L) Recurrent IVL (2 years posthysterectomy) presenting as multilocular cystic masses on ultrasound (I) and MRI (J and K). T2 fat-suppressed imaging reveals minimal low-signal tumor parenchyma on the septa, while T1 fat-suppressed enhanced imaging demonstrates mild-moderate wall enhancement. Histopathology (L, H&E) demonstrates IVL with atypical features: diffuse growth of oval/short spindle-shaped tumor cells with significant interstitial edema and cystic changes, vascular ectasia and extravasation of erythrocytes. Tumor cells exhibit mild atypia, focal mitoses and no necrosis (Immunoprofile: Desmin+, Actin (focal)+, ER/PR+, Ki67∼10%, and CD34 (vascular)+; Calretinin/CD10/S-100/a-inhibin−). Fig 1 dummy alt text (A-D) Transvaginal ultrasound demonstrates characteristic IVL morphologies: beaded (A, arrow), cord-like (B), and mass-like (C-E) pelvic lesions. Color Doppler reveals intralesional and perilesional punctate/short linear vascular signals. Solid-predominant masses display irregular (C, arrows) or crescent-shaped cystic components (D, arrow) within the hypoechoic tumor tissue. (E-H) Multimodal imaging in a case with posthysterectomy IVL recurrence. While ultrasound (E) and CT (F) depict heterogeneous pelvic masses, MRI T2WI (G and H) clearly delineates pathognomonic worm-like morphology (arrows). (I-L) Recurrent IVL (2 years posthysterectomy) presenting as multilocular cystic masses on ultrasound (I) and MRI (J and K). T2 fat-suppressed imaging reveals minimal low-signal tumor parenchyma on the septa, while T1 fat-suppressed enhanced imaging demonstrates mild-moderate wall enhancement. Histopathology (L, H&E) demonstrates IVL with atypical features: diffuse growth of oval/short spindle-shaped tumor cells with significant interstitial edema and cystic changes, vascular ectasia and extravasation of erythrocytes. Tumor cells exhibit mild atypia, focal mitoses and no necrosis (Immunoprofile: Desmin+, Actin (focal)+, ER/PR+, Ki67∼10%, and CD34 (vascular)+; Calretinin/CD10/S-100/a-inhibin−). IVL pelvic lesions were categorized according to their solid-cystic composition: solid (9/19, 47.3%), predominantly solid (7/19, 36.8%), and predominantly cystic (2/19, 10.5%). Solid lesions lacked anechoic areas but demonstrated disorganized intratumoral and peritumoral vascularity on Doppler imaging, typically presenting as punctate or short linear flow signals ( Fig. 1 A, B, and E). Predominantly solid masses featured irregular or crescent-shaped anechoic areas located within or adjacent to the solid components ( Fig. 1 C and D). The predominantly cystic type represented a rare and diagnostically challenging variant, appearing as enlarging multilocular adnexal or pelvic cysts mimicking cystadenomas ( Fig. 1 I); this type is considered an atypical imaging manifestation of IVL. Pathological analysis identified 1 case as atypical IVL with cystic degeneration and significant interstitial edema ( Fig. 1 L), and another as IVL coexisting with endometriosis. Among the 12 patients evaluated by cardiac and abdominal vascular ultrasound, the modality demonstrated a sensitivity of 67.8% (8/12) for detecting abdominal venous or cardiac involvement. Notably, none of the cases with ovarian vein infiltration was identified sonographically. Characteristic sonographic findings included: distension of the venous lumen containing irregular, heterogeneous cord-like hypo- or isoechoic structures; discernible venous flow between the mass and vessel wall without significant hemodynamic compromise; incomplete compressibility under probe pressure; and well-demarcated masses showing dynamic mobility within vascular lumina or cardiac chambers without wall adherence.

Contrast-enhanced CT and/or MRI was performed in 16 patients. Among these, 2 cases showed multilocular cystic morphology, concordant with sonographic findings ( Fig. 1 J and K). The remaining 14 cases demonstrated irregular, heterogeneous soft tissue masses involving the uterus, parametrium, or pelvis cavity. Imaging characteristics included: (1) heterogeneous density on CT ( Fig. 1 F); (2) hypo- to isointensity on T1-weighted MRI and mild hyperintensity on T2-weighted sequences; and (3) mild-to-moderate heterogeneous enhancement. Contrast-enhanced MRI more clearly delineated the characteristic worm-like contours of pelvic masses ( Fig. 1 G and H), a distinctive feature of IVL that is often poorly visualized on ultrasound or CT, which may only show heterogeneous masses ( Fig. 1 E and F). Furthermore, contrast-enhanced MRI demonstrated superior sensitivity in detecting subtle parametrial vascular involvement, clearly revealing intraluminal filling defects ( Fig. 2 A and B)—features often missed on CT and ultrasound. Fig. 2 (A and B) Stage II IVL. MRI-T2WI (A) and enhanced T1WI (B) demonstrate filling defects (arrow) in the tortuous venous lumen of the uterine wall and parametrium. (C) Right ovarian vein involvement in IVL. Coronal CT reconstruction reveals a solid lesion filling the right ovarian vein (arrow), extending into the IVC. (D) Stage III IVL. Enhanced CT axial view shows involvement of the IVC (arrow) and left hepatic vein (short arrow). (E and F) Stage IV IVL. Maximum intensity projection (MIP) image (E) demonstrates involvement of the right pulmonary artery trunk (arrow). The distal pulmonary artery in this case is also affected, presenting as solid nodules in the lung (F, short arrow). Fig 2 dummy alt text (A and B) Stage II IVL. MRI-T2WI (A) and enhanced T1WI (B) demonstrate filling defects (arrow) in the tortuous venous lumen of the uterine wall and parametrium. (C) Right ovarian vein involvement in IVL. Coronal CT reconstruction reveals a solid lesion filling the right ovarian vein (arrow), extending into the IVC. (D) Stage III IVL. Enhanced CT axial view shows involvement of the IVC (arrow) and left hepatic vein (short arrow). (E and F) Stage IV IVL. Maximum intensity projection (MIP) image (E) demonstrates involvement of the right pulmonary artery trunk (arrow). The distal pulmonary artery in this case is also affected, presenting as solid nodules in the lung (F, short arrow). Contrast-enhanced CT and MRI with 3-dimensional (3D) reconstruction clearly delineated 3 distinct vascular extension pathways of IVL: (1) parametrial uterine vein → internal iliac vein → common iliac vein → IVC ( n = 7; Fig. 3 C); (2) left ovarian vein → left renal vein → IVC or right ovarian vein → IVC ( n = 2; Fig. 2 C); and (3) extension via both pathways simultaneously ( n = 9). Fig. 3 (A-C) Stage III IVL (solid). Venous-phase enhanced CT (B) optimally demonstrates continuous filling defects in the IVC (arrow) compared to arterial phase (A). Coronal reconstruction (C) delineates tumor extension along the bilateral parametrial venous plexus → iliac veins → IVC → right atrium. (D-F) Recurrent IVL (sieve-like) with AVM (1-year posthysterectomy). Volume-rendered CT (D) and coronal reconstruction (F) reveal characteristic “loofah sponge” morphology within the IVC. Arterial-phase axial CT (E) reveals a “sieve-like” appearance, with low-density areas representing the lesion (arrow). The residual tumor at the left ovarian vein stump (short arrow) extending to the left renal vein. Fig 3 dummy alt text (A-C) Stage III IVL (solid). Venous-phase enhanced CT (B) optimally demonstrates continuous filling defects in the IVC (arrow) compared to arterial phase (A). Coronal reconstruction (C) delineates tumor extension along the bilateral parametrial venous plexus → iliac veins → IVC → right atrium. (D-F) Recurrent IVL (sieve-like) with AVM (1-year posthysterectomy). Volume-rendered CT (D) and coronal reconstruction (F) reveal characteristic “loofah sponge” morphology within the IVC. Arterial-phase axial CT (E) reveals a “sieve-like” appearance, with low-density areas representing the lesion (arrow). The residual tumor at the left ovarian vein stump (short arrow) extending to the left renal vein. Six cases demonstrated cardiac involvement, and 4 showed pulmonary artery extension. Notably, 1 patient exhibited tumor extension confined to the IVC without cardiac involvement but with bilateral pulmonary metastases, manifesting as multiple well-circumscribed solid nodules. Venous-phase imaging was optimal for identifying intravascular IVL lesions, which appeared as continuous filling defects ( Fig. 3 A-C). In the rare instance of IVL coexisting with a pelvic arteriovenous malformation (AVM), arterial-phase acquisition strikingly demonstrated the venous lesions ( Fig. 3 E) due to early venous enhancement from contrast shunting. Excluding 1 case without parametrial vascular involvement, intravascular lesions manifested as irregular elongated solid densities ( Fig. 3 A-C) in 77.8% of cases (14/18). In the remaining 4 patients (22.2%), axial images demonstrated a “sieve-like” pattern, whereas coronal views revealed characteristic “loofah sponge” morphology ( Fig. 3 D-F). Three-dimensional CT coronal reconstruction clearly delineated anatomical continuity between pelvic masses and IVC lesions, as well as extension to the right atrium, where the lesions appeared thickened and curved, presenting a characteristic “snakehead” or “walking-stick head” appearance ( Fig. 3 C). Additional vascular involvements included: the left hepatic vein in 1 case, showing a well-defined, regular filling defect ( Fig. 2 D); the right pulmonary artery trunk in 1 case ( Fig. 2 E); the distal pulmonary arteries in 3 cases, presenting as multiple well-defined, regular solid nodules in the lower lung lobes, some with visible intraluminal filling defects ( Fig. 2 F); and the renal vein in 3 cases.

The data without potentially identifiable information that support the findings of this study are available on request from the corresponding author.