Centrosome amplification, the presence of more than two centrosomes in a cell is a common feature of most human cancer cell lines. However, little is known about centrosome numbers in human cancers and whether amplification or other numerical aberrations are frequently present. To address this question, we have analyzed a large cohort of primary human epithelial ovarian cancers (EOCs) from 100 patients. We found that rigorous quantitation of centrosome number in tumor samples was extremely challenging due to tumor heterogeneity and extensive tissue disorganization. Interestingly, even if centrosome clusters could be identified, the incidence of centrosome amplification was not comparable to what has been described in cultured cancer cells. Surprisingly, centrosome loss events where a few or many nuclei were not associated with centrosomes were clearly noticed and overall more frequent than centrosome amplification. Our findings highlight the difficulty of characterizing centrosome numbers in human tumors, while revealing a novel paradigm of centrosome number defects in EOCs.

In clinical oncology, only a few applications have been developed using HPV as a personalized tumor marker, a lack most probably related to the limited information obtained by the classical Polymerase Chain Reaction (PCR) approach. To overcome this limitation, we have recently developed the capture-based Next-Generation Sequencing (NGS) “CaptHPV” assay, designed to provide an extensive and comprehensive molecular characterization of HPV DNA sequences associated with neoplasias, ie the sequence of the viral genome (245 genotypes), its physical state, viral load, integration site and genomic alterations at integration locus. These data correspond to highly specific tumor markers that can be used to improve diagnosis and patient’s follow-up. We report here a case that is a straightforward and practical illustration of the power of the CaptHPV method. A patient developed successively a carcinoma of the anal canal and of the tongue. The two tumors were squamous cell carcinoma, found associated with HPV16 using PCR. In order to document a possible metastasis to the tongue from the anal cancer, we performed CaptHPV analysis on the two tumors. The analysis of the anal carcinoma found 55 viral/human hybrid reads allowing the identification of the HPV16 DNA integration in the 4q25 chromosomal band locus with a 178,808 bp deletion in the cell genome. Molecular analysis of the tongue tumor disclosed 6110 reads of HPV16, with a viral pattern strictly identical to that of the anal tumor. A total of 131 hybrid reads between HPV16 and the cell genome were found, corresponding exactly to the same locus of integration of viral DNA at the 4q25 site. The 178,808 bp genomic deletion was also found in the lingual tumor. The exact identity of HPV insertional signatures in the two tumors, demonstrates unambiguously that the tongue tumor derived from the anal cancer whereas neither histological immunophenotyping nor classical viral analysis using PCR could allow a definitive diagnosis. Our observation indicates that the establishment of a detailed cartography of HPV DNA sequences in a tumor specimen provides crucial information for the design of specific biomarkers that can be used for diagnostic, prognostic or predictive purposes.

In HPV-associated carcinomas, some examples of cancer-related genes altered by viral insertion and corresponding to potential therapeutic targets have been described, but no quantitative assessment of these events, including poorly recurrent targets, has been reported to date. To document these occurrences, we built and analyzed a database comprised of 1455 cases, including HPV genotypes and tumor localizations. Host DNA sequences targeted by viral integration were classified as “non-recurrent” (one single reported case; 838 loci), “weakly recurrent” (two reported cases; 82 loci), and highly recurrent (≥3 cases; 43 loci). Whereas the overall rate of cancer-related target genes was 3.3% in the Gencode database, this rate increased to 6.5% in “non-recurrent”, 11.4% in “weakly recurrent”, and 40.1% in “highly recurrent” genes targeted by integration (p = 4.9 × 10⁻⁴). This rate was also significantly higher in tumors associated with high-risk HPV16/18/45 than other genotypes. Among the genes targeted by HPV insertion, 30.2% corresponded to direct or indirect druggable targets, a rate rising to 50% in “highly recurrent” targets. Using data from the literature and the DepMap 23Q4 release database, we found that genes targeted by viral insertion could be new candidates potentially involved in HPV-associated oncogenesis. A more systematic characterization of HPV/host fusion DNA sequences in HPV-associated cancers should provide a better knowledge of HPV-driven carcinogenesis and favor the development of personalize patient treatments.