Background: The significant sex and gender differences that exist in cancer mechanisms, incidence, and survival, have yet to impact clinical practice. One barrier to translation is that cancer phenotypes cannot be segregated into distinct male versus female categories. Instead, within this convenient but contrived dichotomy, male and female cancer phenotypes are highly overlapping and vary between female- and male- skewed extremes. Thus, sex and gender-specific treatments are unrealistic, and our translational goal should be adaptation of treatment to the variable effects of sex and gender on targetable pathways.
Methods: To overcome this obstacle, we profiled the similarities in 8370 transcriptomes of 26 different adult and 4 different pediatric cancer types. We calculated the posterior probabilities of predicting patient sex and gender based on the observed sexes of similar samples in this map of transcriptome similarity.
Results: Transcriptomic index (TI) values were derived from posterior probabilities and allowed us to identify poles with local enrichments for male or female transcriptomes. TI supported deconvolution of transcriptomes into measures of patient-specific activity in sex and gender-biased, targetable pathways. It identified sex and gender-skewed extremes in mechanistic phenotypes like cell cycle signaling and immunity, and precisely positioned each patient's whole transcriptome on an axis of continuously varying sex and gender phenotypes.
Conclusions: Cancer type, patient sex and gender, and TI value provides a novel and patient- specific mechanistic identifier that can be used for realistic sex and gender-adaptations of precision cancer treatment planning.
Keywords: Bayesian analyses; Cancer; Cell cycle regulation; Hallmark pathways; Inflammation/immunity; Personalized medicine; Sex and gender differences.
Some efforts to improve cancer therapy involve the idea of personalizing treatments to who a patient is and how their cancer operates. Personalizing treatment can involve straighforward features like a patient’s age, family cancer history, personal disease and surgical histories, as well as more complex features like analysis of their specific cancer’s mechanisms of growth and spread throughout the body. One glaring omission in common personalization schemes is the sex and gender of the patient. While patient sex and gender is known to substantially affect cancer rates and response to treatment, we do not yet use this information in treatment planning. There are multiple reasons for this but among them is that we tend to think about sex and gender as an either/or categorization. You are either a male/man or a female/woman. This is not accurate as there are many variables that contribute to who an individual is as a male/man or female/woman. This variability is a challenge to incorporating these features into personalized treatment planning. Here, we have developed a method to address this challenge. It is our great hope that this will enable the use of this critically important element of personalization in cancer treatment planning and improve survival rates for all patients.
© 2024. The Author(s).