Computer-enhanced analysis of microvessel density increases the ability to predict cancer stage from prostate needle biopsies, reported David G. Bostwick, MD, of the Mayo Clinic in Rochester, Minnesota.
Sharing with participants at the First Sonoma Conference on Prostate Cancer the interim analysis of data from an ongoing study, Dr. Bostwick noted the study used a single cut-off point for predicting stagewhether prostate cancer was confined to the prostate or had extended outside of the prostate. On the ROC (receiver operator characteristics curve), Dr. Bostwick said, microvessel density analysis increased the accuracy from about .77 to .81. So the incremental increase isnt great, he said. Its significant, but its not fabulous.
Is the increase in predictive value enough to justify the use of microvessel density analysis? Im not sure thats the case, Dr. Bostwick claimed. Im delivering a message of possible utility for selected patients.
Microvessel density may be useful for treatment decisionsperhaps watchful waiting versus radiation versus perhaps some other therapies for individual patients. He does not advocate its use to stratify patients for surgery vs no surgery.
186 Needle Biopsies Are Evaluated
Five institutions are participating in the study, including the Mayo Clinic in Rochester, Minnesota, where Dr. Bostwick is affiliated with the Department of Laboratory Medicine and Pathology.
Researchers evaluated 186 randomly selected needle biopsies and matched totally embedded radical prostatectomy samplesthe gold standard for stage, said Dr. Bostwickwith preoperative PSA concentration and patient demographics. Gleason score and optimized microvessel density were determined from the needle biopsy samples.
Although staging was initially performed at each institution, it was later verified by independent review of the prostatectomy samples. In 12% of the cases, the reviewers could not agree on the stage, and the cases were excluded.
For research purposes only, full or complete sampling was performed on the study specimens. Dr. Bostwick acknowledged that this is not currently the standard of care in pathology. The standard of care is reasonable partial sampling of the prostate.
Digital Image Analysis System Used
An automated digital image analysis system measured microvessel morphology and calculated the optimized microvessel density in the biopsies. Previous studies had suggested several links (summarized below) between increased microvessel density and extraprostatic spread of adenocarcinoma.
There is a significant increase in microvessel density in prostatic intraepi-thelial neoplasia and carcinoma when compared with normal prostatic tissue.
- Mean blood vessel count is higher in tumors with metastases than in those without metastases, and most, but not all, studies demonstrate a correlation with pathologic stage
- Microvessel density appears to be an independent predictor of cancer progression.
- Increased microvessel density in prostate cancer is probably related to the production of angiogenesis-associated growth, similar to other organs.
Computer Accentuates Blood Vessels
Using slides showing side-by-side comparisons of benign and malignant tissue, Dr. Bostwick pointed out how the blood vessels in benign tissue tend to be looser and have a lower density.
There may also be large, peripheral sinuses adjacent to the acini. In cancer, of course, the acini are much more compact, he reminded the conference participants, and the blood vessels are very small. Its like a Medusas head of blood vessels within the prostate.
In a videotape presentation, Dr. Bostwick pointed out how the computer rendering showed blood vessels forming an intricate mixture of vessels that are acutely branching, as well as mixing throughout...So, we have a compact gland, and this proliferation of small blood vessels. Which came first? Did the blood vessels allow the cancer to grow at that site, or did the cancer induce the blood vessels? We think that the cancer induced the blood vessels.
The computer has an advantage over the human eye, Dr. Bostwick said. It can actuallyusing contrast enhancement or reductionaccentuate the blood vessels themselves, and then subtract them out. So you can actually count the number of vessels, the size of the vessels, the density of the vessels, and the roundness of the vessels. Many different things can be done.
The computer program also separates out the empty space. So we look at true blood vessel versus stroma, Dr. Bostwick noted, not blood vessel versus stroma and empty space in the prostate. And theres a lot of it, because the prostate is very spongy.
Some studies have also shown that microvessel density analysis has the advantage of being relatively uniform, from area to area, in the prostate, Dr. Bostwick added. Within the tumor itself, the center may have a little bit more in the way of blood vessels than the edge. But the variation is not that great. There may be up to 40% to 50% difference from the center to the edge. In multiple foci, similarly, it appears that the variation is within a reasonable limit. Thats only been shown in a single, unpublished study, but it looks reasonably good.
Dr. Bostwick said that its now estimated that 85% of patients with prostate cancer have multiple focal cancer and that the average is 2.6 cancers. Men dont have prostate cancer, Dr. Bostwick stated. They have prostate cancers.
Finding Useful Prognostic Factors
The microvessel density study represents yet another attempt to determine the optimal combination of techniques for predicting which patients have prostate cancers that will progress and which do not.
You can look at prognosis in a number of different ways, Dr. Bostwick said. These include the characteristics of the patient, the extent or architecture of the tumor, the biology of the tumor, and treatment and treatment response.
Microvessel density probably falls into the patient characteristicthat is the host response to the tumorbut you could argue that its also part of the tumor biology, he said.
To be useful, Dr. Bostwick said that prognostic factors should be:
- Significant, independent, or at least strongly complementary;
- Validated by clinical testing;
- Able to be measured in multiple centers;
- Interpretable by the clinician; and
- Have therapeutic implications.
Finding new and better prognostic indicators is important, he noted, because current clinical staging
of prostate cancer is often inaccurate, requiring upstaging upon final pathologic review.