Researchers investigate prognostic role of diversity of chromatin compartments in gynecological cancers

A novel marker might contribute to selecting high-risk stage I ovarian carcinoma patients for adjuvant chemotherapy and low-risk endometrial carcinoma patients for less extensive surgery.

Chromatin organization affects gene expression and contributes to the development of cancer. It has previously been shown that automatic quantification of chromatin heterogeneity can be applied to identify patients with increased risk of cancer recurrence and death in several cancer types.

Institute for Cancer Genetics and Informatics (ICGI) is a department at Oslo University Hospital. A joint team of researchers, developers and lab-personnel have for more than 15 years been developing methods that can improve cancer diagnostics through combining biomedicine and informatics. In recent years, we have focused on utilizing new technology, often referred to as machine learning or artificial intelligence (AI).

Researchers and collaborators of the institute recently published a paper in Cancers, investigating the prognostic role of diversity of chromatin compartments in gynecological carcinomas. The diversity was quantified using the entropy of chromatin compartment sizes and optical densities within compartments. The prognostic value of the novel marker was related to chromatin heterogeneity and pathological risk classifications.

According to the papers' first author Andreas Kleppe, the study indicates that more direct analysis of chromatin compartments in cancer cell nuclei might provide a more accurate prognostic marker than purely statistical analyses of chromatin heterogeneity.

Markers based on analysis of chromatin organization in cancer cell nuclei appear to improve the prediction of recurrence and death from gynecological cancer beyond what is possible using markers currently in use in clinical practice. This suggests that we can use the analysis of chromatin organization to tailor the treatment of patients with gynecological cancer. This might reduce overtreatment of patients with good prognosis and allow treatment of patients with poor prognosis to be intensified, says Kleppe.

Pan-cancer prognostic markers

In 2018, researchers at the Institute for Cancer Genetics and Informatics at Oslo University Hospital and associates used statistical texture analysis of cancer cell nuclei stained for DNA to develop a pan-cancer prognostic marker of chromatin heterogeneity. The study was published in The Lancet Oncology and showed the potential of machine learning algorithms to estimate prognosis across established cancers.

Machine learning algorithms analyzed the chromatin organization in 461 000 images of tumor cell nuclei stained for DNA from 390 patients (discovery cohort) treated for stage I or II colorectal cancer at the Aker University Hospital (Oslo, Norway). The resulting marker of chromatin heterogeneity, termed Nucleotyping, was subsequently independently validated in six patient cohorts: 442 patients with stage I or II colorectal cancer in the Gloucester Colorectal Cancer Study (UK); 391 patients with stage II colorectal cancer in the QUASAR 2 trial; 246 patients with stage I ovarian carcinoma; 354 patients with uterine sarcoma; 307 patients with prostate carcinoma; and 791 patients with endometrial carcinoma. In all cohorts, chromatin heterogeneous tumors were associated with worse cancer-specific survival.

The same year, Birgitte Nielsen, Kleppe and colleagues applied machine learning algorithms to detect single nuclei with high chromatin entropy in gynecological cancers. The study was published in the Journal of the National Cancer Institute and found that most patients had less than 25% nuclei with high chromatin entropy. However, the 10-15% of the patients with a higher proportion of nuclei with high chromatin entropy had significantly more recurrences and cancer deaths.

Analysis of chromatin compartments

The new study builds upon these findings and reports analyses of two cohorts consisting of 1037 patients with gynecological carcinoma. The prognostic value of the diversity of chromatin compartments was moderately strongly correlated with the prognostic value of chromatin heterogeneity. The novel marker supplemented established clinical and pathological markers, adding prognostic information that identified patients with more than twice the hazard of developing cancer recurrence and death.

Integrating the novel marker with pathological risk classifications gave three risk groups with distinctly different prognoses. This combined marker might provide a way to more appropriately select high-risk stage I ovarian carcinoma patients for adjuvant chemotherapy and preoperatively identify low-risk endometrial carcinoma patients who are candidates for less extensive surgery.

Predicts recurrence and death

Kleppe says the results add to previous findings that have indicated that analyses of chromatin organization in cancer cell nuclei can predict recurrence and death from gynecological cancer.

New in this study is that the chromatin reorganization is linked more directly to chromatin compartments. In particular, the increased number and size of weakly and highly condensed chromatin compartments indicates a worse prognosis.

Dr Andreas Kleppe is a researcher at ICGI and an adjunct associate professor at the Department of Informatics at the University of Oslo. His research interests include the application of machine learning to cancer diagnostics and prognostics.