Estimating and comparing cancer progression risks under varying surveillance protocols

Abstract

Outcomes after cancer diagnosis and treatment are often observed at discrete times via doctor-patient encounters or specialized diagnostic examinations. Despite their ubiquity as endpoints in cancer studies, such outcomes pose challenges for analysis. In particular, comparisons between studies or patient populations with different surveillance schema may be confounded by differences in visit frequencies. We present a statistical framework based on multistate and hidden Markov models that represents events on a continuous time scale given data with discrete observation times. To demonstrate this framework, we consider the problem of comparing risks of prostate cancer progression across multiple active surveillance cohorts with different surveillance frequencies. We show that the different surveillance schedules partially explain observed differences in the progression risks between cohorts. Our application permits the conclusion that differences in underlying cancer progression risks across cohorts persist after accounting for different surveillance frequencies.

Prostate cancer mortality and metastasis under different biopsy frequencies in North American active surveillance cohorts

“…Our model of underlying upgrading from GS6 to GS7 based on the 4 cohorts has been described previously . The estimation approach accommodates imperfect biopsy sensitivity and specificity in detecting GS7 disease and includes age at diagnosis, baseline prostate‐specific antigen (PSA), and PSA velocity as predictors of upgrading risk.…”

“…We generated 4 different populations (N = 50,000 each) that reflected the baseline distribution of age and PSA characteristics and the underlying risk of upgrading in the 4 North American cohorts . The number 50,000 was selected to minimize Monte Carlo error in quantities of interest but allow for computational feasibility.…”

“…For each individual, we simulated a time of underlying upgrade, given cohort, age at diagnosis, PSA intercept, and slope. We then superimposed a range of biopsy schedules (Table ) on the disease histories. For each schedule, we determined the time at which the biopsy detected the underlying upgrade, assuming a biopsy sensitivity of 60% for detecting GS7 disease …”

“…There are several challenges involved in using existing data for comparing AS protocols. First, patient populations may vary in underlying risk . Second, there is limited information on downstream outcomes, such as metastases and deaths, because of the low risk of progression and lengthy follow‐up needed to accumulate an adequate number of events.…”

“…Our goal is to produce results that reflect the range of risks encountered in contemporary cohorts, enabling us to ascertain the robustness of our findings to differences in risk of underlying upgrade. Those included in the current study are from the multi‐institutional Canary Prostate Active Surveillance Study (PASS) and cohorts from the University of California at San Francisco (UCSF), JHU, and UT . Underlying upgrading is defined as the first point at which a biopsy with perfect sensitivity to detect grade progression would detect a GS ≥7 in a man who was initially diagnosed with GS6 disease and chose AS.…”