Monday, August 13, 2018

Salvage Radiation Dose: Decision-Making Under Uncertainty

A large, well-done, confirmed randomized clinical trial (RCT) is the only evidence that proves that one therapy is better than another. According to current consensus, this is deemed "Level 1a" evidence. But this high level of evidence is seldom available. This is especially true of prostate cancer because it takes so long to achieve acceptable endpoints like overall survival, prostate cancer-specific survival, and metastasis-free survival. Such studies are very expensive and difficult to carry out.

Alexidis et al. analyzed the National Cancer Database for men treated with adjuvant or salvage radiation therapy (SRT) after prostatectomy failure from 2003 to 2012. SRT with doses above 66.6 Gy were labeled "high dose," and SRT with doses above 70.2 Gy were labeled "very high dose." Between 2003 and 2012:

  • High dose SRT utilization increased from 30% to 64%
  • Very high dose SRT utilization increased from 5% to 11%
  • Utilization of high and very high dose rates was greatest at academic centers, lowest at community centers.

The authors decry the fact that this doubling of high dose SRT took place in the absence of RCTs that would definitively establish proof. They point out that the evidence for it is based on observational studies (see, for example, King and Kapp and Ohri et al.), which are fraught with confounding due to stage migration,  selection bias and ascertainment bias. Stage migration was the result of better imaging becoming increasingly available to rule out SRT from patients already harboring occult distant metastases. Also, three randomized clinical trials published in the middle of the observational period convinced many radiation oncologists that earlier SRT led to better tumor control than waiting. Selection bias occurred because the patients selected to get higher doses of radiation were healthier and those whose cancer was less progressed -- they would have done better regardless of the dose. Ascertainment bias resulted from the longer observational period for patients treated in 2003 vs. 2012 - the opportunity for treatment failure increases with the amount of time that has passed. The authors also doubt that biochemical recurrence-free survival (which is what was used in observational studies) is a good enough surrogate endpoint for overall survival. They are right that all these factors may be confounding the previous retrospective analyses, and the only way to know with certainty is to conduct a trial where patients are randomized to receive high or low SRT doses,  and follow patients long enough so that median survival or at least metastasis-free survival is reached in the low dose group.

There has been one randomized clinical trial of SRT dose escalation in the modern era. The SAKK 09/10 trial found little difference in acute toxicity symptoms at 70 Gy compared to 64 Gy, but patient-reported urinary symptoms worsened. Unfortunately, many patients were treated with three-dimensional conformal radiation therapy (3D-CRT), which had higher toxicity than the IMRT in widespread use now. Also, it uses freedom from biochemical failure (not yet reported) as its surrogate endpoint.

So, what is a patient to do in the absence of Level 1a evidence? Should he accept the higher doses with possibly added toxicity and better tumor control, or should he go for a lower dose with possibly less toxicity and less tumor control?

As a compromise, Mantini et al. recently reported 5-year biochemical disease-free survival (bDFS) and other outcomes for patients who received higher dose SRT (70.2 Gy vs. 64.8 Gy) depending on their post-operative pathology. They also may have received (depending on pathology) whole pelvic radiation and adjuvant hormone therapy. Those patients who received the higher dose had equivalent 5-yr bDFS in spite of their worse disease characteristics. Those who received only 64.8 Gy still had a 5-year bDFS as high as 92%. We do not know how many of those recurrent men with favorable disease characteristics actually needed any SRT. They were all treated with 3D-CRT and toxicity was not reported.

The other thing we can do when our information is imperfect is go through the Bradford Hill checklist. It can give us more confidence if we have to make a decision based on less than Level 1 evidence. The factors that ought to be considered are:

  • Strength of Association (larger associations are more likely (but not necessarily) causal)
  • Consistency of Data (independent studies all lead to the same conclusion)
  • Specificity (a very specific population is differentially affected)
  • Temporality (the effect has to occur after the cause)
  • Biological gradient (too some extent, more drug/radiation dose leads to more effect) 
  • Plausibility (one can come up with a plausible explanation)
  • Coherence (lab studies demonstrate a plausible mechanism for the observed effect)
  • Experiment (has the effect been prevented by modifying the cause)
  • Analogy (similar factors may be considered)


Unfortunately, the authors did not refer to Dr. King's more recent analysis of SRT dose/response, which we discussed in depth here. He looked at 71 studies, demonstrating consistency. While it is not Level 1 evidence, it is Level 2a evidence. In it, he observes that the salvage radiation dose response conforms exactly to the primary radiation dose response.  In other words, the prostate tumor is equally radio-resistant whether it is in the prostate or the prostate bed. This increases the plausibility of a dose effect of SRT. What's more, dose escalation was proven to be beneficial for biochemical recurrence-free survival, metastasis-free survival, and freedom from lifelong ADT use, for primary radiation in intermediate risk men by a RCT (RTOG 0126). So, we also have greater confidence in SRT dose escalation by analogy.

RTOG 0126 did not find an increase with higher dose in 8-year overall survival or cancer-specific survival. This calls into question whether these longer-term effects are really useful endpoints if we are to be able to obtain and use the results of any clinical trial in a reasonable time frame.

Dr. King proposed a randomized clinical trial of 76 Gy vs. 66 Gy for SRT. Meanwhile, he is routinely giving his SRT patients at UCLA 72 Gy. Dr. Zelefsky at Memorial Sloan Kettering Cancer Center and other eminent radiation oncologists have also upped the radiation dose to 72 Gy. Such doses seem to be safe and effective, but it is one of many factors in the SRT treatment decision that must be carefully considered by patients and their doctors.