The small increase in high-grade cancers is due to ascertainment bias introduced by the effect of 5ARIs. This bias is due to two factors: gland cytoreduction and improved performance of the PSA. The re-analysis of pathology using the modified Gleason score in fact introduced two sources of variation from the original data: the effect of the modified Gleason scoring system, and the effect of inter-rater reliability. This was reflected in the reduced number of Gleason 7 tumors in the re-analysis. If 5ARIs increased high-grade cancer (i.e., Gleason pattern 4–5), one would have expected an increase in Gleason 7 disease. A number of modeling studies have sought to estimate the effect of gland cytoreduction and improved PSA performance. Correcting for ascertainment bias, these have estimated a reduction in all cancers of 39% and a reduction of high-grade cancer of 16%. These studies were not accepted by the FDA. In a summary of the decision, Theoret calculated that, adjusting for change in prostate volume, the number of Gleason 8–10 cancers was increased by a factor of 1.5. This analysis assumes that volume reduction is uniform between peripheral zone and transition zone. In fact, many of these patients likely harboured anterior cancers. Recent data (not reviewed by the FDA) has shown that, amongst men with an elevated PSA and persistent negative biopsies, a substantial proportion harbour anterior cancers. These are often high grade (Margel et al, J Urol, in press). Anterior cancers in large prostates are notoriously difficult to target with TRUS guided biopsies. It is likely that the increase in high-grade cancer reflects nothing more than greater likelihood of detecting these cancers once the glands have been reduced in size by 5ARI treatment. In fact, the greater likelihood of finding high-grade anterior cancer may be a benefit of 5ARI. Further, the FDA discounted the benefit of avoiding a diagnosis of low-grade prostate cancer. Another source of imbalance in the REDUCE trial relates to the two sets of biopsies mandated, at year 2 and year 4. There was no difference in high-grade cancer at the year two biopsy. However, during the first round of biopsies, 142 more patients had cancers detected in the placebo arm, likely because they grew more rapidly than those on Dutasteride. These patients were taken off study. Assuming that both groups had the same number of cancers at study entry, this means that 142 more patients in the Dutasteride group had cancers going in to the 2nd round of biopsies at year 4. Assuming that 8% of these cancers were in fact Gleason 8–10, one would expect approximately 12 more Gleason 8–10 cancers identified in the Dutasteride group on the 2nd set of biopsies; exactly what was found (12 vs 1, 0.5% vs .04%). Further, despite the Dutasteride group harbouring more cancers going into the 2nd biopsy, and the improved performance of PSA in men on Dutasteride, the number of cases identified was significantly fewer. It is also likely that the men with high-grade cancer would have been identified and treated in a timely fashion. Using the National Comprehensive Cancer Network (NCCN) guideline for biopsy (NCCN Clinical Practice Guidelines, 2011, www.nccn.org), ¾ of the high-grade cancer cases would have been diagnosed. There was no evidence that Dutasteride interfered with the detection of high-grade cancers using PSA. Economic analysis of the benefit of 5ARI in prevention suggests that the cost per cancer avoided is reasonable, and in line with other accepted prevention strategies. Therefore we believe the data from PCPT and REDUCE justifies the use of 5ARIs in prostate cancer prevention. This is particularly true given the other benefits of the drug in the prevention of the sequelae of BPH progression. |
