Is overall survival a useful endpoint for evaluating therapies for intermediate risk patients?

In a recent commentary, we looked at the utility of surrogate endpoints in evaluating therapies. In an abstract presented at ASCO, Malouf et al. examined the large National Cancer Data Base to determine whether there is an association between the use of brachytherapy (BT), external beam radiation (EBRT), or a combination of both (CT) and overall survival in intermediate risk patients.

They found records on 122,405 patients treated from 2004-2013 who were staged IIA. IIA is an AJCC risk category that is similar to the NCCN intermediate risk category, except that it excludes those clinically staged with cancer in both lobes (stage T2c). The average age of the patients at diagnosis were:

• ·      EBRT: 69 years of age
• ·      BT and CT: 66 years of age

The study provides no information about the radiation doses used.

The average survival, and the percent who survived 10 years were:

• ·      EBRT: 109 months, 61.5%
• ·      BT: 116 months, 72.9%
• ·      CT: 116 months, 73.1%

Survival differences were statistically significant between EBRT cohort and those who received the two other therapies.

The authors conclude:

“The method of radiotherapy used contributes to the survival of patients with stage IIA prostate cancer, with brachytherapy with or without EBRT having improved survival. Careful selection of the proper treatment regimen should be used.”

Now, when we look at US actuarial tables, we see the expected survival for a 66 year-old man is 16.93 years (203 months), and 14.81 years for a 69 year-old man (177 months). So the men treated with EBRT should have lived 26 months less; yet they lived only 7 months less – a relative survival gain for some unknown reason. It is also unknown why overall survival in both cohorts was so much less than actuarial expectations.

Using the Memorial Sloan Kettering nomogram for life expectancy where intermediate risk cancer has been diagnosed but not yet treated, and assuming no significant co-morbidities or risk factors, and allowing only for the difference in age, the expected 10-year survival statistics for untreated prostate cancer are as follows:

Among the 66 year old men (BT and CT cohorts):

• ·      71% would still be alive, which is close to the observed 73% among those who were treated
• ·      20% would have died of other causes
• ·      9% would have died of prostate cancer

Among the 69 year old men (EBRT cohort):

• ·      67% would still be alive, which is somewhat higher than the observed 63% among those who were treated
• ·      25% would have died of other causes
• ·      8% would have died of prostate cancer

What we learn from this is that for a man who has a life expectancy of ten years or less, watchful waiting may be a better choice than radical treatment.

We see that it is impossible to attribute the difference in the overall survival to prostate cancer, let alone to any of the treatments received. What we needed to know is prostate cancer-specific mortality, and we have no idea from their analysis how, if at all, it was affected. Because of the very low rate of prostate cancer-specific mortality at 10 years, even in untreated patients, it takes a very long time to be able to detect differences in the efficacy of various treatments based on this endpoint; hence, the importance of surrogate endpoints. The authors’ conclusions are completely unfounded based on the data they presented.