Use of mpMRI and PSMA PET/CT to aid in salvage radiation decision-making

Because the success or failure of salvage radiation (SRT) hinges upon whether micrometastases are already systemic at the time of treatment, evidence that the cancer is still local improves the odds that SRT will be successful.. One way of finding local tumors is to use multiparametric MRI (mpMRI). mpMRI can detect tumors down to about a limit of 4 mm, and may be able to find tumors even when their PSA output is low.

Sharma et al. at the Mayo Clinic retrospectively examined the records of 473 men who were treated with SRT and who had an mpMRI prior to treatment from 2003 to 2013. Among men with a pre-treatment PSA ≤ 0.5 ng/ml, 5-year biochemical failure was:

• 39% among those with a negative mpMRI
• 12% among those with a positive mpMRI

Adding mpMRI to the updated Stephenson nomogram (see this link) increased its predictive accuracy for PSA recurrence after SRT from 71% to 77%. Perhaps its accuracy would increase even further if the MRI was confirmed by a biopsy of the suspicious tissue to eliminate any false positives.

Like the detection of a positive margin in post-prostatectomy pathology, detection of a local tumor using mpMRI increases the probability that SRT will be successful. Although the radiation dose to the suspicious lesion can be boosted (see this link), it is unknown whether such a boost actually increases efficacy when the entire prostate bed is adequately treated. It is also unknown what effect it might have on toxicity. Moreover, it is hard to argue for a reduced dose elsewhere in the prostate bed because of the known limitation of mpMRI in detecting smaller tumors, and the multi-focal nature of prostate cancer spreading.


Emmett et al. at St. Vincent Hospital in Sydney performed a Ga-68-PSMA-11 PET/CT on 164 men with rising PSA (PSA range: 0.05-1.0 ng/ml) after prostatectomy who received SRT. After eliminating patients who also had systemic therapy, there were 140 evaluable patients. They had a pre-SRT PSA of 0.23 (interquartile range 0.14-0.35).  As expected, detection rates went up with increasing PSA;

• <0.2 ng/ml: 50%
• 0.20-0.29 ng/ml: 64%
• 0.30-0.39 ng/ml: 67%
• ≥0.40 ng/ml: 81% 

They only had 10.5 months of median follow-up, and defined a favorable PSA response to SRT as a decrease of at least 50% in PSA and a PSA ≤ 0.1 ng/ml (those receiving adjuvant ADT were eliminated from the follow-up PSA-response analysis). The results should be interpreted with caution because of the very short follow up and low sample sizes. A short-term PSA response only indicates local control, and may not endure if systemic micrometastases were present.

PET/CT was negative in 38% (62/164). 45% of those men (27/60) had SRT to the prostate bed, and 7/27 had SRT to the pelvic lymph nodes field too. In the "negative" detection group, 86% had a favorable PSA response to SRT. Unfortunately, more than half of the PET-negative men never received SRT. This should serve as a caution against over-reliance on PET/CT. PET/CT is not good at detecting micrometastases in the prostate bed. The prostate bed is also a difficult place to detect PSMA-avid cancer because of masking from urinary excretion. We also know little about the natural history of PSMA development in prostate cancer -- it  may very well be that earlier forms of the cancer that may not express PSMA may be most vulnerable to SRT. SRT should never be withheld from an area based solely on negative PSMA findings.

PET/CT was positive in the prostate bed only in 23% (38/164). All of them had SRT to the prostate bed, and 17/36 had SRT to the pelvic lymph node field too. In the "prostate-bed only" detection group, 81% had a favorable PSA response to SRT. Recent evidence indicates that pelvic lymph node SRT increases effectiveness (see this link). Radiation of the pelvic lymph nodes should be considered in spite of negative nodal PSMA findings.

PET/CT was positive in pelvic lymph nodes in 25% (41/164). 87% (26/30) of them had SRT to the prostate bed and to the targeted pelvic lymph nodes. In the "pelvic lymph node" detection group, 61.5% had a favorable PSA response to SRT. The entire pelvic lymph node field and not just isolated lymph nodes should receive SRT for the reasons stated above.

PET/CT was positive for distant metastases in 14% (23/164). Nevertheless, 60% (10/15) of them had SRT to the prostate bed (and, I suppose, to the entire pelvic lymph node field), and 6/10 had metastasis-directed SBRT too. In the "distant metastasis" detection group, only 30% had a favorable PSA response to SRT. Only 1 of the 6 who had metastasis-directed SBRT had a favorable PSA response. When there are known distant metastases, treatment of the prostate bed, pelvic lymph nodes, and of metastases remains a controversial treatment.

The PET/CT was a better predictor of SRT response than PSA, Gleason score, stage, or surgical margin status. The most valuable finding of this small, short-term analysis was that metastases can sometimes be detected at fairly low PSA (as low as 0.1 ng/ml), and it may be possible to rule out SRT in those cases. Conversely, when distant metastases cannot be detected, SRT success rates may be very good.

We will require longer follow-up, larger sample size, prospective studies to establish the utility of mpMRI and PSMA PET/CT in SRT decision making. The two imaging techniques are complementary - the MRI is not as PSA-dependent and is not masked by the urinary excretion of the radiotracer, while the PET scan is highly specific for cancer. Both are useless in detecting tumors with a dimension smaller than 4 mm, so it would be a mistake to think that what is detected is all there is.