The Prostate Cancer Intervention Versus Observation Trial (PIVOT)

The Prostate Cancer Intervention Versus Observation Trial (PIVOT) should prove interest
ing in that the study design will permit observation of the natural history of a potentially lethal malignant disease, influenced only by palliative treatments. My comments will focus on the concerns raised by this study design. I will not address possible biases of the trial introduced by: (1) enrollment of less than 20% of the eligible population; (2) an enrollment rate per participating center of less than 3 patients per year; (3) a 7-year enrollment period; and (4) a 12-year follow-up (for a total trial duration of 19 years).

The natural history of malignant tumors is characterized by growth by repetitive cell division at distant sites. The aim of traditional treatment has been to eradicate the primary tumor by excisional or ablative therapy. Observation (or expectant) treatment schema logically permit malignant cells to replicate at a pace dictated by their inherent biology. If prostate-specific antigen (PSA) levels reflect the tumor volume, then the doubling time of prostatic tumors is approximately 18 months, with a range of 12 to 24 months.

There is also concern that, over time, malignant tumors show a progressive loss of cell differentiation. Serial biopsy data from observed Scandinavian patients demonstrate a shift from diploid to uniploid cells and/or an increase in Gleason sum in approximately 5% of the residual population each year.[1] Our data also show a statistically significant increase in Gleason sum with advancing age.

Cancer-Specific Survival Data

Our data indicate that cancer-specific survival is a function of the extent of disease, regardless of Gleason sum, PSA level at the time of treatment, or other factors (Figure 1). Small-volume tumors, even those with high Gleason sums, can be salvaged by surgical intervention if the disease is organ- or specimen-confined (Figure 2).

Furthermore, our data show that PSA levels at the time of biopsy bear a direct relationship to the extent of disease (or the number of patients who are potentially curable). It seems logical to assume, therefore, that denying patients a proactive treatment option (through random assignment to observation) will shift these patients, with time, to an extended-disease category. In other words, their disease will behave as if it were margin-positive. As all surgically treated margin-positive patients who die from prostate cancer are exposed to androgen deprivation prior to death (and yet experience a 55% cancer-specific death rate at 15 years), it seems reasonable to predict that the patients assigned to observation will approach this figure, even assuming that all of them have organ-confined disease at entry and all progress to margin-positive disease in 3 to 5 years after diagnosis.

I hope that my pessimistic projections prove incorrect. Certainly, all of the observed population will die with prostate cancer. If competing risks of death are adequately controlled, however, members of the observed population will also be at risk of dying from prostate cancer.