PSA After Radiation for Prostate Cancer

In this paper, Dr. Kuban et al address controversies surrounding the use of posttreatment prostatespecific antigen (PSA) in determining outcome after radiotherapy. They base most of their discussion on their own observations of prostate cancer outcomes in more than 4,000 patients following external-beam radiotherapy alone.[1,2] I had the privilege of writing an editorial on their earlier companion papers, and I made the argument then that although some definitions were slightly better than the American Society for Therapeutic Radiology and Oncology (ASTRO) definition, the differences were not impressive enough to recommend changing the standard for determining outcome after external- beam radiotherapy.[3] Dr. Kuban and colleagues raise the provocative idea that perhaps we could create "a universal definition of failure that might be applied to multiple modalities, including surgery." Rather than writing another editorial congratulating these investigators for "a job well done" and making the point that "switching to a new [definition] may not be worthwhile," I will focus on the question of whether we can ever have a "universal definition." In fact, some investigators have assumed that the ASTRO definition should be applied to patients treated with either external-beam irradiation or brachytherapy, implying that a universal definition might already exist.[ 4,5] Carrying this further, other investigators have gone so far as to apply the ASTRO definition to patients treated with radical prostatectomy.[6] Is this attempt to use the ASTRO definition universally appropriate, or should we adopt a more sensitive definition? Universal Application of the ASTRO Definition
We already have a definition of PSA failure after radical prostatectomy (0.2 ng/mL) that is likely to be far more sensitive and specific than the ASTRO definition, so why not use it after external-beam radiotherapy? As Dr. Kuban and coauthors demonstrate, the specificity of a failure definition based on a PSA level of 0.2 ng/mL after external-beam irradiation was only 9%. This is not surprising because the day-to-day variability of PSA values in men with an intact prostate exceeds the thresholds currently accepted as highly predictive of recurrence after surgery. Thus, although the level to which the PSA declines correlates roughly with the risk of PSA failure after treatment, a surgical definition is too strict for patients treated with external-beam radiotherapy. Some investigators have applied the ASTRO definition to patients with permanent prostate implants (PPI), but this approach has not been studied using the statistical approaches described by Thames et al for patients treated with external-beam irradiation. Although Dr. Kuban's data suggest that the least reliable definition of PSA failure following external-beam radiotherapy involves the use of threshold values > 0.2 ng/mL, this may very well not be true for patients treated with PPI.[2] Because of the ablative nature of the high doses associated with this modality, the specificity of a strict definition is likely to be much better when applied to PPI-based therapy. Five years after a patient receives external-beam irradiation to 45 Gy plus PPI irradiation to 108 Gy, a persistent PSA level > 1 ng/mL is likely to reflect persistent disease. However, after external-beam radiotherapy to 70 Gy, such a patient is more likely to have residual benign glands that can produce PSA. What about applying the ASTRO definition to patients after prostatectomy? Such an application might work if urologists were willing to accept ultrasensitive assays with a detectable value (eg, PSA ≥ 0.02 ng/mL) that rises two or three times consecutively.[ 7] Given that the threshold for defining PSA failure after prostatectomy has been rising, it seems doubtful that the urology community will do an about-face and head toward an ultrasensitive test.[8] I would argue that they cannot have it both ways, ie, require a minimal value > 0.2 ng/mL or 0.4 ng/mL and then count failure only after a required number of increases. Many men with their prostates in place have PSAs in that range. How can urologists justify defining a patient as "cured of prostate cancer" if he is walking around with a PSA well into the normal range? Perhaps the most promising approach for developing a universal definition (in patients not receiving hormonal therapy), would be to use the slope of the PSA over a specified recent period of time.[9] With this approach, bouncing PSA values can be "self-neutralized." This occurs because upward bounces are offset by downward bounces, thus eliminating the noise associated with random PSA fluctuations while capturing consistent PSA progression even if the increases are not consecutive. Conclusions
Unfortunately, the problem is far more complex than Dr. Kuban or I have addressed thus far. This is due to the fact that patients who are most likely to have a rising PSA are those at greatest risk of recurrence at the time of treatment. Many, if not all of these patients, should have received hormonal therapy as part of their initial therapy. Although it is less problematic to compare groups of patients receiving hormonal therapy for the same duration, it is more problematic to define recurrences using PSA if the duration is not standardized.[10] For this reason, I would argue that we are a long way away from developing a universal definition. PSA values are extremely sensitive to hormonal manipulations, and there is great variability between patients in the rate at which they recover their testosterone, which is complicated by variations in the duration of hormonal therapy. Thus, I would doubt that one size will ever fit all.[11] To define PSA failure in this setting, I consider the patient's risk of recurrence, his testosterone level, how far out he is, his age, and his pretreatment PSA. On some occasions, I also consider his prior history of prostatitis, the presence of local symptoms and all of his PSA trends, and then I use my own judgment to decide whether I think his disease has recurred. The bottom line is, we are not close to developing a truly "universal definition" that can replace good judgment. In other words, although I can't define it for every type of treatment, "I know it when I see it."