Modes of biochemical failure after primary radiation therapy may identify aggressive sub-types.

The most widely used definition of biochemical failure after primary radiation therapy is the so-called “Phoenix” definition. It’s called that because it was adopted by consensus at a meeting of the American Society for Radiation Oncology (ASTRO) at a meeting in Phoenix, AZ in 2005. It defined biochemical failure as a rise of PSA of 2.0 ng/ml over the lowest PSA recorded after radiation – the nadir. The major reason for the change was because the earlier definition – the “ASTRO” definition -identified many post-treatment PSA patterns as biochemical failures when they were really only temporary rises or bounces.

The older definition of biochemical failure, the ASTRO definition, was adopted by consensus in 1997 and had two parts:

A1: three consecutive rises in post-treatment PSA after nadir was reached, or

A2: the initiation of salvage androgen deprivation therapy (ADT) for any reason other than A1.

The former definition is still important because (1) clinical trials that were completed between 1997 and 2005 used the “ASTRO” definition, and we have received a wealth of useful long-term results off of them, (2) similar biochemical failure patterns continue to occur alongside the “Phoenix” definition, and (3) we can learn something about how radiation failures may occur from the PSA patterns observed in those studies, and the risk associated with each of those patterns of failure.

There were four PSA failure patterns observed other than the three consecutive PSA rises (A1). Those A2 failure patterns were:

Pattern 1: PSA never decreased after radiation – no nadir.

Pattern 2: PSA increased, but not 3 consecutive increases

Pattern 3: PSA changed in an irregular manner (other than Pattern 2)

Pattern 4: Patients in whom distant metastases were detected soon after radiation therapy

Hamstra et al. analyzed the post-radiation biochemical failure patterns of two major randomized clinical trials – RTOG 9202 and RTOG 9413.  Across the two studies, 2,799 patients were observed, and 1181 (42%) experienced biochemical failure. Among the biochemical failures:

• ·      56% were A1 failures (3 consecutive rises)
• ·      44% were A2 failures. Among those:

o   10% were Pattern 1 failures

o   55% were Pattern 2 failures

o   12% were Pattern 3 failures

o   23% were Pattern 4 failures

• ·      There were no differences in age, pre-treatment PSA, stage, Gleason score, or lymph node status between A1 and A2 failures.

The table below shows the metastasis rate, the all-cause mortality and local failure rate after five years. We see that:

• ·      A2 failures were associated with worse outcomes compared to A1 failures.
• ·      The local failure rate did not differ significantly.
• ·      Pattern 4 (early metastases) and Pattern 1 (failure to reach nadir) had the worst outcomes.
• ·      Pattern 2 failures, the most prevalent A2 type, had a similar all-cause mortality and a lower metastasis rate compared to A1 failures.

	5-year metastasis rate (percent)	5-year all-cause mortality (percent)	5-year local failure rate (percent)
A1 failures	16	12	20
A2 failures (total)	29	26	21
Pattern 1	37	49	24
Pattern 2	8	10	21
Pattern 3	7	36	22
Pattern 4	87	47	20

The apparent lower risk associated with A1 failure supports the hypothesis that initiating salvage hormone therapy may represent overtreatment in many of those patients.

The authors note that A2 failures may represent biologically more aggressive types of prostate cancer not otherwise identified by Gleason score, PSA, stage, or nodal status, and are worth further study. It would be interesting to run a comparative genomic analysis to see if there are identifiable phenotypes that are prognostic for A2 failure.

note: Thanks to Dr. Daniel Hamstra for providing us with a copy of the full text of the recent paper by him and his colleagues.