No abstract

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Background: Prostate cancer is the leading cancer in men in sub-Saharan Africa (SSA) regarding incidence and mortality. Published data from a few registries in SSA suggest that the rates are still rising, but there is little comprehensive information on the time trends of prostate cancer incidence. Methods: We analyzed registry data on 13,170 incident prostate cancer cases in men aged 40 years or above, from 12 population-based cancer registries in 11 SSA countries, with at least a 10-year time span of comparable data. Results: We observed an increase in cumulative risks (CR) and age-standardized incidence rates (ASR) over time in all registries (statistically significant in all but one). The highest values of CR were found in Seychelles and Harare (Zimbabwe). The highest annual increase in the ASRs was seen in Seychelles and Eastern Cape (South Africa), whereas the lowest was seen in Mauritius. We mainly found a steady increase in incidence with age and during successive periods. Conclusions: This analysis reveals that prostate cancer incidence rates are rising in many populations in SSA—often very rapidly—which is in contrast to recent observations worldwide. We acknowledge that the reasons are multifactorial and largely remain unclear, but believe that they are primarily associated with improvements in health care systems, for example, a broader use of prostate-specific antigen testing. Impact: This study is the first to compare population-level data on time trends of prostate cancer incidence between multiple countries of SSA, presenting the different rates of increase in 11 of them.

BACKGROUND: Although prostate cancer (PCa) is the most commonly diagnosed cancer in men of sub-Saharan Africa (SSA), little is known about its management and survival. The objective of the current study was to describe the presentation, patterns of diagnosis, treatment, and survival of patients with PCa in 10 countries of SSA. METHODS: In this observational registry study with data collection from 2010 to 2018, the authors drew a random sample of 738 patients with PCa who were registered in 11 population-based cancer registries. They described proportions of patients receiving recommended care and presented survival estimates. Multivariable Cox regression was used to calculate hazard ratios comparing the survival of patients with and without cancer-directed therapies (CDTs). RESULTS:The study included 693 patients, and tumor characteristics and treatment information were available for 365 patients, 37.3% of whom had metastatic disease. Only 11.2% had a complete diagnostic workup for risk stratification. Among the nonmetastatic patients, 17.5% received curative-intent therapy, and 27.5% received no CDT. Among the metastatic patients, 59.6% received androgen deprivation therapy. The 3-and 5-year age-standardized relative survival for 491 patients with survival time information was 58.8% (95% confidence interval [CI], 48.5%-67.7%) and 56.9% (95% CI, 39.8%-70.9%), respectively. In a multivariable analysis, survival was considerably poorer among patients without CDT versus those with therapy. CONCLUSIONS: This study shows that a large proportion of patients with PCa in SSA are not staged or are insufficiently staged and undertreated, and this results in unfavorable survival. These findings reemphasize the need for improving diagnostic workup and access to care in SSA in order to mitigate the heavy burden of the disease in the region.

SummaryNon‐Hodgkin lymphoma (NHL) is the sixth most common cancer in Sub‐Saharan Africa (SSA). Comprehensive diagnostics of NHL are essential for effective treatment. Our objective was to assess the frequency of NHL subtypes, disease stage and further diagnostic aspects. Eleven population‐based cancer registries in 10 countries participated in our observational study. A random sample of 516 patients was included. Histological confirmation of NHL was available for 76.2% and cytological confirmation for another 17.3%. NHL subclassification was determined in 42.1%. Of these, diffuse large B cell lymphoma, chronic lymphocytic leukaemia and Burkitt lymphoma were the most common subtypes identified (48.8%, 18.4% and 6.0%, respectively). We traced 293 patients, for whom recorded data were amended using clinical records. For these, information on stage, human immunodeficiency virus (HIV) status and Eastern Cooperative Oncology Group Performance Status (ECOG PS) was available for 60.8%, 52.6% and 45.1%, respectively. Stage at diagnosis was advanced for 130 of 178 (73.0%) patients, HIV status was positive for 97 of 154 (63.0%) and ECOG PS was ≥2 for 81 of 132 (61.4%). Knowledge about NHL subclassification and baseline clinical characteristics is crucial for guideline‐recommended treatment. Hence, regionally adapted investments in pathological capacity, as well as standardised clinical diagnostics, will significantly improve the therapeutic precision for NHL in SSA.

Background. Cervical cancer (CC) is the most common female cancer in many countries of sub-Saharan Africa (SSA). We assessed treatment guideline adherence and its association with overall survival (OS). Methods. Our observational study covered nine populationbased cancer registries in eight countries: Benin, Ethiopia, Ivory Coast, Kenya, Mali, Mozambique, Uganda, and Zimbabwe. Random samples of 44-125 patients diagnosed from 2010 to 2016 were selected in each. Cancer-directed therapy (CDT) was evaluated for degree of adherence to National Comprehensive Cancer Network (U.S.) Guidelines. Results. Of 632 patients, 15.8% received CDT with curative potential: 5.2% guideline-adherent, 2.4% with minor deviations, and 8.2% with major deviations. CDT was not documented or was without curative potential in 22%; 15.7% were diagnosed with International Federation of Gynecology and Obstetrics (FIGO) stage IV disease. Adherence was not assessed in 46.9% (no stage or follow-up documented, 11.9%, or records not traced, 35.1%). The largest share of guideline-adherent CDT was observed in Nairobi (49%) and the smallest in Maputo (4%). In patients with FIGO stage I-III disease (n = 190), minor and major guideline deviations were associated with impaired OS (hazard rate ratio [HRR], 1.73; 95% confidence interval [CI], 0.36-8.37; HRR, 1.97; CI, 0.59-6.56, respectively). CDT without curative potential (HRR, 3.88; CI,) and no CDT (HRR, 9.43; CI, 3.03-29.33) showed substantially worse survival. Conclusion.We found that only one in six patients with cervical cancer in SSA received CDT with curative potential. At least one-fifth and possibly up to two-thirds of women never accessed CDT, despite curable disease, resulting in impaired OS. Investments into more radiotherapy, chemotherapy, and surgical training could change the fatal outcomes of many patients. The Oncologist 2021;25:1-10 Implications for Practice: Despite evidence-based interventions including guideline-adherent treatment for cervical cancer (CC), there is huge disparity in survival across the globe. This comprehensive multinational population-based registry study

The histopathological phenotype of tumors reflects the underlying genetic makeup. Deep learning can predict genetic alterations from tissue morphology, but it is unclear how well these predictions generalize to external datasets. Here, we present a deep learning pipeline based on self-supervised feature extraction which achieves a robust predictability of genetic alterations in two large multicentric datasets of seven tumor types.

The histopathological phenotype of tumors reflects the underlying genetic makeup. Deep learning can predict genetic alterations from pathology slides, but it is unclear how well these predictions generalize to external datasets. We performed a systematic study on Deep-Learning-based prediction of genetic alterations from histology, using two large datasets of multiple tumor types. We show that an analysis pipeline that integrates self-supervised feature extraction and attention-based multiple instance learning achieves a robust predictability and generalizability.