Many emerging tools for comprehensive molecular profiling of malignant lesions demand fresh frozen tissue with a high
tumor purity. Often, a
tumor epithelial content of at least 80% is recommended. This approach may lead to a systematic bias, and therefore we explore if this introduces a selection of cases with a certain phenotype in
cervical cancer. Clinicopathologic data for a population-based cohort of 328 patients have been studied. Fresh frozen
tumor specimens were available for 151 of these patients and investigated for epithelial
tumor cell portion in
hematoxylin-stained frozen sections by light microscopy. The estimated
tumor purity in the samples was compared with FIGO stage, histopathologic characteristics and survival. High
tumor purity was significantly more often found in
squamous cell carcinoma (SCC) compared to
adenocarcinoma (AC) (P=0.03). For the subgroup of AC (n=40), there was a significant association between high
tumor purity in the fresh frozen samples and later occurrence of recurrent disease (P=0.04). In SCC, no significant associations between
tumor purity and disease stage, grade or outcome were found. Apparently in line with this, grade was found to influence prognosis in AC, but not in SCC. Our findings suggest that selection of samples based on high
tumor purity in fresh frozen tissue may introduce a selection bias toward aggressive disease for the subgroup of AC, but not for SCC of the cervix. Thus, the prevalence of potential molecular
biomarkers identified in AC in particular should be validated in a population-based setting to further explore clinical relevance. Also, molecular
biomarkers only prevalent in subgroups with low
tumor purity may go undetected in sample collections enriched for high
tumor purity.