5 Discussion

This study demonstrates that the extent of omental sampling has a measurable impact on the detection of microscopic metastatic disease in ovarian and endometrial carcinomas. Our findings support and extend previous model-based observations suggesting that limited sampling may be inadequate for detecting small metastatic foci, particularly in grossly normal-appearing omentum.

Significant variation exists among previous studies and pathology protocols regarding omental sampling strategies. Although some retrospective analyses have reported that submission of one or two blocks from the omentum yields metastasis detection rates similar to more extensive sampling^1,2, these findings may have been influenced by cohort characteristics, tumor biology, and inclusion of cases with macroscopically evident disease. In contrast, studies specifically focusing on staging-negative or macroscopically normal omentum have emphasized that limited sampling may miss microscopic metastatic disease³. Indeed, one model study demonstrated that increasing the block number from five to ten raised the sensitivity for detecting microscopic metastases from 82% to 95%⁴.

Our findings indicate that microscopic metastases can be optimally sampled by obtaining at least 4 samples from grossly normal-appearing omental specimens, achieving >95% detection sensitivity. Increasing this number further to 7 cassettes approaches 99.6% detection. Notably, our 4-cassette threshold for >95% sensitivity differs from the model study by Skala and Hagemann, which required 10 blocks to achieve 95% sensitivity. This discrepancy likely reflects differences in our cohort composition, micrometastasis definition (<0.5 cm), and the fact that our estimates are derived from a larger dataset of real-world cases rather than a simulation model. Our finding shows greater concordance with recommendations for more extensive sampling and provides real-world data from a large retrospective series from routine pathology practice.

The detection of microscopic omental metastases even in cases without macroscopic omental abnormality reveals the limitations of macroscopic evaluation alone. This is particularly important in endometrial carcinoma, as the presence of omental involvement even at the microscopic level results in classification of the disease as FIGO stage IVB and is associated with significant prognostic adversity. Given the strong association of omental metastasis with other poor prognostic features, inadequate omental sampling may lead to disease understaging and potentially suboptimal clinical management.

The variation in current literature recommendations, ranging from two to four blocks to five to ten blocks, reveals the absence of true consensus in this area and that current recommendations are largely based on non-outcome-based data. Our study’s findings suggest that increasing omental sampling, particularly in staging scenarios where other metastatic sites are negative, may enhance diagnostic accuracy. These results indicate that routine pathologic evaluation and sampling strategies in gynecologic malignancies need to be reconsidered.

5.1 Strengths and Limitations

Strengths of our study include the large patient cohort (n=1098), inclusion of both ovarian and endometrial carcinomas, and focus on a practical question frequently encountered in daily pathology practice. Our findings provide important data aimed at bridging the gap between guideline recommendations and real-world practice. Importantly, all statistical values presented are computed directly from the data with complete reproducibility.

Limitations include the retrospective design and lack of long-term follow-up data. An inherent limitation of any retrospective sampling adequacy study is verification bias: we can only analyze cases in which metastasis was detected. Patients with true microscopic metastasis that was entirely missed by the original sampling protocol are classified as metastasis-negative and excluded from analysis. This means our detection probability estimates are conditional on at least one cassette being positive, and the true rate of occult omental metastasis may be higher than the 4.2% observed in this cohort. The observation that 76.1% of cases had tumor in multiple blocks might suggest dependent sampling; however, since the spatial relationship between sampled blocks was not recorded, we cannot determine whether this reflects clustered or diffuse disease. Regardless of the spatial pattern, the multi-block positivity reflects the extent of disease rather than sampling dependency. In a macroscopically negative omentum, the pathologist cannot distinguish tumor-bearing from tumor-free tissue, so each cassette represents an operationally independent sample. This distinguishes our model from scenarios where a pathologist might target suspicious areas. Additionally, the beta-binomial sensitivity analysis confirmed that the 4-cassette recommendation remains robust even after accounting for inter-case heterogeneity. Prospective studies with standardized sampling protocols would provide more definitive evidence.

5.2 Clinical Implications

In conclusion, this study demonstrates that the extent of omental sampling significantly affects the detection of microscopic metastatic disease in ovarian and endometrial carcinomas. Particularly in cases with grossly normal omentum, increasing the sampling extent can improve staging accuracy and play a decisive role in clinical decision-making processes.

Our data suggest that submission of at least 4 cassettes from grossly normal-appearing omentum achieves >95% sensitivity for detecting microscopic metastases. This recommendation balances the need for accurate staging with practical considerations of tissue processing and pathology workload. Future prospective studies should validate these findings and determine whether this sampling strategy impacts clinical outcomes and treatment decisions.

# Discussion ```{r setup-functions} #| include: false # Source manuscript functions source("./R/manuscript_functions.R") source("./R/data_for_manuscript.R") ``` This study demonstrates that the extent of omental sampling has a measurable impact on the detection of microscopic metastatic disease in ovarian and endometrial carcinomas. Our findings support and extend previous model-based observations suggesting that limited sampling may be inadequate for detecting small metastatic foci, particularly in grossly normal-appearing omentum. Significant variation exists among previous studies and pathology protocols regarding omental sampling strategies. Although some retrospective analyses have reported that submission of one or two blocks from the omentum yields metastasis detection rates similar to more extensive sampling [@thomas2019pathologic; @gilbert2012sampling], these findings may have been influenced by cohort characteristics, tumor biology, and inclusion of cases with macroscopically evident disease. In contrast, studies specifically focusing on staging-negative or macroscopically normal omentum have emphasized that limited sampling may miss microscopic metastatic disease [@usubütün2019omentectomy]. Indeed, one model study demonstrated that increasing the block number from five to ten raised the sensitivity for detecting microscopic metastases from 82% to 95% [@usubütün2024optimal]. Our findings indicate that microscopic metastases can be optimally sampled by obtaining at least **4 samples** from grossly normal-appearing omental specimens, achieving **>95% detection sensitivity**. Increasing this number further to 7 cassettes approaches `r cumulative_detection$cumulative_pct[7]`% detection. Notably, our 4-cassette threshold for >95% sensitivity differs from the model study by Skala and Hagemann, which required 10 blocks to achieve 95% sensitivity. This discrepancy likely reflects differences in our cohort composition, micrometastasis definition (<0.5 cm), and the fact that our estimates are derived from a larger dataset of real-world cases rather than a simulation model. Our finding shows greater concordance with recommendations for more extensive sampling and provides real-world data from a large retrospective series from routine pathology practice. The detection of microscopic omental metastases even in cases without macroscopic omental abnormality reveals the limitations of macroscopic evaluation alone. This is particularly important in endometrial carcinoma, as the presence of omental involvement even at the microscopic level results in classification of the disease as FIGO stage IVB and is associated with significant prognostic adversity. Given the strong association of omental metastasis with other poor prognostic features, inadequate omental sampling may lead to disease understaging and potentially suboptimal clinical management. The variation in current literature recommendations, ranging from two to four blocks to five to ten blocks, reveals the absence of true consensus in this area and that current recommendations are largely based on non-outcome-based data. Our study's findings suggest that increasing omental sampling, particularly in staging scenarios where other metastatic sites are negative, may enhance diagnostic accuracy. These results indicate that routine pathologic evaluation and sampling strategies in gynecologic malignancies need to be reconsidered. ## Strengths and Limitations Strengths of our study include the large patient cohort (n=`r cohort_stats$n_total`), inclusion of both ovarian and endometrial carcinomas, and focus on a practical question frequently encountered in daily pathology practice. Our findings provide important data aimed at bridging the gap between guideline recommendations and real-world practice. Importantly, all statistical values presented are computed directly from the data with complete reproducibility. Limitations include the retrospective design and lack of long-term follow-up data. An inherent limitation of any retrospective sampling adequacy study is verification bias: we can only analyze cases in which metastasis was detected. Patients with true microscopic metastasis that was entirely missed by the original sampling protocol are classified as metastasis-negative and excluded from analysis. This means our detection probability estimates are conditional on at least one cassette being positive, and the true rate of occult omental metastasis may be higher than the `r metastasis_stats$pct_micro_only`% observed in this cohort. The observation that `r clustering_stats$pct_multiple_blocks`% of cases had tumor in multiple blocks might suggest dependent sampling; however, since the spatial relationship between sampled blocks was not recorded, we cannot determine whether this reflects clustered or diffuse disease. Regardless of the spatial pattern, the multi-block positivity reflects the extent of disease rather than sampling dependency. In a macroscopically negative omentum, the pathologist cannot distinguish tumor-bearing from tumor-free tissue, so each cassette represents an operationally independent sample. This distinguishes our model from scenarios where a pathologist might target suspicious areas. Additionally, the beta-binomial sensitivity analysis confirmed that the 4-cassette recommendation remains robust even after accounting for inter-case heterogeneity. Prospective studies with standardized sampling protocols would provide more definitive evidence. ## Clinical Implications In conclusion, this study demonstrates that the extent of omental sampling significantly affects the detection of microscopic metastatic disease in ovarian and endometrial carcinomas. Particularly in cases with grossly normal omentum, increasing the sampling extent can improve staging accuracy and play a decisive role in clinical decision-making processes. **Our data suggest that submission of at least 4 cassettes from grossly normal-appearing omentum achieves >95% sensitivity for detecting microscopic metastases.** This recommendation balances the need for accurate staging with practical considerations of tissue processing and pathology workload. Future prospective studies should validate these findings and determine whether this sampling strategy impacts clinical outcomes and treatment decisions.