[Paper Review] Statistical interpretations and new findings on Variation in Cancer Risk Among Tissues

Robert Noble, Oliver Kaltz+1 more|arXiv (Cornell University)|Feb 3, 2015

Cancer Genomics and Diagnostics21 references3 citations

TL;DR

This study reanalyzes Tomasetti and Vogelstein's (2015) data on cancer risk variation by separating the independent effects of stem cell number (tissue size) and stem cell division rate. It finds that both factors independently explain significant variation in cancer incidence, with a plateau in risk at ~0.6% when excluding carcinogen- and disease-associated cancers, and that anatomical site explains most remaining variation, suggesting tissue-specific protective mechanisms influenced by evolution.

ABSTRACT

Tomasetti and Vogelstein (2015a) find that the incidence of a set of cancer types is correlated with the total number of normal stem cell divisions. Here, we separate the effects of standing stem cell number (i.e., organ or tissue size) and per stem cell lifetime replication rate. We show that each has a statistically significant and independent effect on explaining variation in cancer incidence over the 31 cases considered by Tomasetti and Vogelstein. When considering the total number of stem cell divisions and when removing cases associated with disease or carcinogens, we find that cancer incidence attains a plateau of approximately 0.6% incidence for the cases considered by these authors. We further demonstrate that grouping by anatomical site explains most of the remaining variation in risk between cancer types. This new analysis suggests that cancer risk depends not only on the number of stem cell divisions but varies enormously ($\\sim$10,000 times) depending on the stem cell's environment. Future research should investigate how tissue characteristics (anatomical site, type, size, stem cell divisions) explain cancer incidence over a wider range of cancers, to what extent different tissues express specific protective mechanisms, and whether any differential protection can be attributed to natural selection.

Motivation & Objective

• To disentangle the independent contributions of tissue size (stem cell number) and stem cell division rate to cancer incidence variation.
• To assess whether the strong correlation between total stem cell divisions and cancer risk masks underlying biological factors such as tissue environment and protective mechanisms.
• To test whether anatomical site explains residual variation in cancer risk after accounting for stem cell number and division rate.
• To evaluate the implications of these findings for evolutionary theories of cancer suppression and natural selection in human tissues.

Proposed method

• Conducted multiple regression analysis with cancer incidence as the response variable, using log-transformed stem cell number (s) and lifetime replication rate per stem cell (d) as predictors.
• Performed partial regression to isolate the unique contribution of d after controlling for s, and vice versa, using adjusted R² values to quantify explained variance.
• Used a plateau model to test whether cancer incidence saturates at high levels of total stem cell divisions, fitting the data to a nonlinear function.
• Grouped cancers by anatomical site (e.g., pancreas, bone, intestine) to assess site-specific risk per stem cell division.
• Applied statistical corrections to isolate the effects of s and d, using residual analysis from regression models.
• Evaluated the Extra Risk Score (ERS) and compared it with a simplified classification of cancers likely caused by carcinogens, to assess consistency with original findings.

Experimental results

Research questions

• RQ1 To what extent do stem cell number and division rate independently explain variation in cancer incidence across tissues?
• RQ2 Does cancer incidence plateau at a specific level when only endogenous, replication-driven cancers are considered?
• RQ3 How much of the remaining variation in cancer risk is explained by anatomical site after accounting for stem cell number and division rate?
• RQ4 Can tissue-specific protective mechanisms explain the observed differences in cancer risk per stem cell division across anatomical sites?
• RQ5 To what extent might natural selection have shaped tissue-specific cancer suppression mechanisms in humans?

Key findings

• Stem cell number (tissue size) explains approximately 50% of the variation in cancer incidence, independent of division rate.
• After controlling for stem cell number, division rate still explains a statistically significant and independent portion of risk variation.
• Cancer incidence plateaus at approximately 0.6% for cancers not caused by disease or carcinogens when considering total stem cell divisions.
• Anatomical site explains most of the remaining variation in cancer risk, with each site showing a distinct risk per stem cell division.
• The risk per stem cell division varies by up to 10,000-fold across different anatomical sites, indicating strong tissue-specific protective mechanisms.
• The data suggest that natural selection may have shaped tissue-specific cancer suppression, particularly in tissues with high total stem cell divisions.