Tumor-Stroma Proportion May Predict Clinical Outcomes in High-Grade Serous Ovarian Cancer


Progression-free survival and platinum chemoresistance may be predicted by tumor-stroma proportion in patients with high-grade serous ovarian cancer.

Tumor-Stroma Proportion May Predict Clinical Outcomes in High-Grade Serous Ovarian Cancer

Tumor-Stroma Proportion May Predict Clinical Outcomes in High-Grade Serous Ovarian Cancer

Tumor-stroma proportion (TSP) may predict clinical outcome measures, including progression-free survival (PFS), platinum chemoresistance, and overall survival (OS), in high-grade serous ovarian cancer, according to findings from a retrospective analysis that were published in JAMA Network Open.1

This prognostic study included tumors from 103 patients with ovarian carcinoma collected by The Cancer Genome Atlas (TCGA) that were diagnosed and treated at several institutions from 1993 to 2013; tumors from an independent cohort of resected diagnostic and tissue microarray clinical specimens from 192 patients with suspected or known high-grade serous ovarian carcinomas collected from 2004 to 2014 by the University of Tübingen in Germany were also assessed. No significant association between TSP levels and chemoresistance was observed in the TCGA cohort. In the Tübingen cohort, more patients with high TSP (34.1%; n = 14) were platinum resistant vs those with low TSP (14.6%; n = 22). Conversely, more patients with low TSP (85.4%; n = 129) were platinum sensitive vs those with high TSP (65.9%; n = 27).

TSP, defined as the extent of tumor involvement with stromatous tumor microenvironment components, has been associated with poor outcomes in several cancers.2 Furthermore, a prospective observational study published in 2019 demonstrated that 4 out of 5 newly diagnosed, platinum-resistant ovarian tumors harboring TSP exhibited a high TSP, in contrast to 5 out of 19 platinum-sensitive tumors (= .047).3

“The discovery that high TSP was associated with platinum resistance offered TSP as a potentially effective and low-cost predictive biomarker of drug resistance that, if confirmed and validated, could advance the field of ovarian cancer treatment by helping to tailor treatment for patients in whom platinum therapy was predicted to be less effective than in others,” Emil Lou, MD, PhD, FACP, lead study author, and coauthors, wrote in the paper.1 “Thus, we next sought to confirm these results in a larger patient cohort.”

Lou is a medical oncologist, neuro-oncologist, and gastrointestinal oncologist, as well as an associate professor of medicine in the Division of Hematology, Oncology, and Transplantation at the University of Minnesota Medical School in Minneapolis.

This study was conducted from January 2021 to January 2024. The primary outcomes were the predictive ability of TSP for platinum chemoresistance, PFS, and OS.

TSP scoring was conducted by 1 clinician and 1 pathologist who identified the most densely populated cellular areas of each tumor specimen slide and labeled the relative amounts of tissue represented by cancer cells vs stromal tissue using 50% as a previously validated cutoff. Tumors with a stroma population of less than 50% were considered to have low TSP, and those with a stroma population of 50% or greater were considered to have high TSP. Stromal tissue that did not clearly correlate with the peritumoral stroma was excluded from the evaluation, and areas that were necrotic, empty, mucinous, or large and inflammatory were not included in TSP scoring.

Patients in the TCGA cohort had a mean age of 61.6 years (SD, 11.1), 70.9% had International Federation of Gynecology and Obstetrics (FIGO) stage III disease, and 73.8% had residual disease after surgical debulking. Furthermore, 11.7% and 30.1% of patients had chemoresistant disease and unknown chemoresistance status, respectively, which limited confirmation in this analysis. A total of 18 patients (17.5%) had high TSP, and the remaining 85 patients had low TSP. TSP levels were balanced across clinical and demographic characteristics.

Patients in the Tübingen cohort had a mean age of 63.7 years (SD, 11.1) at diagnosis, 69.8% had FIGO stage III disease, and 49.5% had node-positive disease. Moreover, 68.8% of patients had no evidence of distant metastasis, and 56.3% had residual disease after surgical debulking. Forty-one patients (21.4%) had high TSP, and 151 patients (78.6%) had low TSP. Fourteen patients (34.1%) with high TSP had FIGO stage IV disease at the time of initial diagnosis vs 22 patients (14.6%) with low TSP (= .04). In both the TSP-high and -low groups, the most frequent T and N stages were T3C and N0, respectively; The authors noted that this may be explained by the higher rate of distant metastases in the high TSP group (34.1%) vs the low TSP group (14.6%; P = .01). TSP levels were balanced across age, tumor size, tumor number, and extent of affected lymph nodes.

All patients in the TCGA cohort underwent surgery followed by standard-of-care (SOC) adjuvant chemotherapy; no history of neoadjuvant chemotherapy was reported in this population. All patients in the Tübingen cohort underwent debulking surgical resection of their tumors followed by SOC systemic therapy.

In the TGCA cohort, high TSP following initial diagnosis and treatment was not associated with a worse PFS rate (HR, 1.661; 95% CI, 0.7661-3.603; = .20). Additionally, high TSP after the first several years from diagnosis was not associated with OS (HR, 1.906; 95% CI, 0.9622-3.776; = .06). The authors noted that since chemoresistance status was unknown or not listed for 31 patients in this cohort, further investigation using a larger and more complete dataset is needed to determine whether there is an association between high TSP and a worse PFS rate.

In the Tübingen cohort, when adjusted for primary metastasis, age at diagnosis, and the presence of residual disease, high TSP was associated with significantly lower PFS (HR, 1.586; 95% CI, 1.093-2.302; = .02) and OS (HR, 1.867; 95% CI, 1.249-2.789; = .002) rates.

Patients in the Tübingen cohort with low TSP were more likely to achieve PFS vs those with high TSP at all time points during follow-up (HR, 1.573; 95% CI, 1.090-2.270; = .02). The presence of residual disease following surgery was also associated with worse PFS rates (HR, 2.038; 95% CI, 1.436-2.892; < .001). However, the presence of distant metastases, which were more commonly observed in patients with high TSP, was not associated with PFS rates after adjusting for the contributions of other variables. Additionally, patients with low TSP had a higher OS rate at all time points excluding the initial months (HR, 0.726; 95% CI, 1.166-2.557; = .006). The presence of residual disease (HR, 2.604; 95% CI, 1.764-3.844; < .001), metastases (HR, 1.572; 95% CI, 1.040-2.375; = .03), and age (HR, 1.035; 95% CI, 1.018-1.052; < .001) were also associated with a lower OS rate.

Since several retrospective studies investigating reactive tumor stroma patterns use a tissue microarray (TMA)–based approach with available resection specimens, the investigators aimed to determine whether the TMA approach would generate the same results as whole-slide assessment, as differences in tumor preparation could lead to potential outcome differences. In this study, TMAs consisted of 6 0.8-mm diameter punches for each clinical specimen. Investigators assigned a single TSP category to each patient depending on the mode from their 6 TSP values. From this process, patients were classified as stroma-rich (high TSP) or stroma-poor (low TSP). A univariable analysis was conducted to compare PFS and OS rates for patients with high vs low TSP.

An analysis of TMAs from 185 patients from the Tübingen cohort, included 143 patients who were TSP low and 42 patients who were TSP high, translating to a 15.6% disparity in TSP levels between the whole-slide assessment and the TMA-based scoring. This univariable analysis showed an association between high TSP and significantly lower PFS (HR, 1.675; 95% CI, 1.012-2.772; = .04) and OS (HR, 2.491; 95% CI, 1.585-3.912; < .001) rates.

The investigators also used the Tübingen cohort to validate their 2019 findings that high TSP is associated with the emergence of resistance to platinum chemotherapy in patients with ovarian cancer. After adjusting for primary metastasis, age, and the presence of residual disease, this analysis found that patients with chemoresistant tumors were significantly more likely to have high TSP vs those with chemosensitive tumors (OR, 2.861; 95% CI, 1.256-6.515; = .01). The estimated area under the receiver operating characteristic curve for the model that predicted platinum chemotherapy resistance based on TSP, metastasis, age at diagnosis, the presence of residual disease, and lymph node spread was 0.7644 (95% CI, 0.639-0.889).

“Taken together, these results not only suggest that chemoresistance in patients with [high] TSP is associated with poorer outcomes also observed in these patients, but also suggest that TSP may have specificity and sensitivity high enough to be used in the clinical setting to predict which patients are going to respond to treatments or benefit from newer therapeutic options,” the authors emphasized.

The authors noted that limitations of this study include those inherent to using a retrospective dataset, as well as the inexact nature of assessing residual disease after attempted maximal cytoreduction and debulking and the manual process of assessing slides for TSP. Additionally, the small sample sizes of some of the patient subsets and the intratumoral heterogeneity of ovarian cancer warrant further evaluation of TSP levels in prospective validation studies.

“Here, we demonstrate that TSP serves as an additional stratifying factor for determining cases at highest risk of recurrence and death that are likely due in part to lack of efficacy of standard platinum-based chemotherapies. With this confirmation, we propose that TSP should be further standardized and incorporated into prospective clinical trials as a correlative predictive biomarker for drug resistance,” the authors concluded.


  1. Lou E, Clemente V, Grube M, et al. Tumor-stroma proportion to predict chemoresistance in patients with ovarian cancer. JAMA Netw Open. 2024;7(2):e240407.2024 Feb 5. doi:10.1001/jamanetworkopen.2024.0407
  2. Micke P, Strell C, Mattsson J, et al. The prognostic impact of the tumour stroma fraction: a machine learning-based analysis in 16 human solid tumour types. EBioMedicine. 2021;65:103269. doi:10.1016/j.ebiom.2021.103269
  3. Lou E, Vogel RI, Hoostal S, et al. Tumor-stroma proportion as a predictive biomarker of resistance to platinum-based chemotherapy in patients with ovarian cancer. JAMA Oncol. 2019;5(8):1222-1224. doi:10.1001/jamaoncol.2019.1943
Related Content
© 2024 MJH Life Sciences

All rights reserved.