OAR

King et al. said...
Interpreting rare events in the clinical trial of perirectal spacer gel for men with prostate cancer receiving IMRT: resisting the draw to arrive at conclusions.

Christopher R. King PhD, MD, Jason Wang PhD, Patrick Kupelian MD, and Michael L. Steinberg MD
Dept. of Radiation Oncology
UCLA

We read with great interest the results of this study (1).

The era of dose escalation with IMRT has led to major developments in technique including image-guidance via fiducials or CBCT. These have led to significant improvements in toxicity that are now thankfully infrequent and low grade, and have also led to major improvements in clinical outcomes. With the evolution of prostate radiotherapy furthered by SBRT, these techniques have become essential. In a never-ending quest for improvement in outcomes and reduction of toxicity, one approach has been to temporarily alter a patient’s internal anatomy to favor rectal dose sparing. This approach was used in the aforementioned multi-institutional study (funded by Augmenix, Inc., the manufacturer of SpaceOAR gel) by injecting absorbable polyethylene glycol hydrogel in the perirectal space between the prostate and rectal wall. All patients received fiducial-based image-guided IMRT 79.2 Gy in 1.8-Gy fractions.

The authors’ only statistically significant clinical finding is quoted as: ‘A significant reduction in late (3-15 months) rectal toxicity severity in the spacer group was observed (p=0.04), with a 2.0% and 7.0% late rectal toxicity incidence in the spacer and control groups, respectively.’

Reported in their Table 3, the raw data on late rectal toxicity (CTCAE) are the following:

Toxicity Spacer Control
Grade 0 145 66 p=0.044
Grade 1 3 4
Grade ≥2 0 1

For this particular analysis the authors unfortunately chose to use the Cochran-Mantel-Haenszel test. This statistical test is defined as appropriate when examining categorical outcomes that are repeated under different conditions, with that condition also being tested as one of the co-factors (2, 3). This is not the case here since all patients are pooled together in a single table for analysis and no accounting of confounders are made. The appropriate statistic for this type of data is either Fisher’s exact test, or the generalized Chi-square test. When we plug their numbers into these tests we obtain p=0.096 for Fisher’s exact test, and p=0.125 for Chi-square (or p=0.55 when using Yate’s correction which is required in this case since events are below threshold). So in our view the author’s results are in fact not statistically meaningful.

However, even if one accepts their raw data at face value, ignoring for the moment statistical analysis, then the difference comes down to a single patient event for grade 1 toxicity and a single patient event for grade 2. In other words, one would need to inject 74 patients with this gel in order to potentially benefit 1 patient from experiencing grade 1 rectal toxicity and maybe up to 1 patient with grade 2 toxicity.

Since the authors have not accounted for other confounders of radiotherapy-related rectal toxicity (use of anticoagulants, history of hemorrhoids, diet, baseline bowel habits, etc.) it is possible that one single outlier patient could account for the observations. Regardless of one’s perspective of interpretation, the data are hardly compelling or convincing.

Lastly, we do not feel that the authors have rigorously analyzed the EPIC QOL trends. They have chosen to report only the proportion of patients who experienced a 5-point or a 10-point difference in the 100-point EPIC scale over time. While this is qualitatively informative, the rigorous method of statistical analysis of EPIC scores has been established as follows: a clinically relevant change in the QOL is defined as a difference from baseline to follow-up that exceeds half a standard deviation of the baseline variance (4, 5). Without this analysis it is not possible to say whether the differences observed here are meaningful. Furthermore, one is not able to gauge whether these differences are relevant to overall QOL or patient satisfaction, or whether these small differences will converge over follow-up that is longer than the study’s 15 months.

We applaud the authors' efforts for exploring this interesting concept in a controlled trial which tested a means by which toxicity of IMRT (or SBRT) could be further improved. However, we urge a deeper objective examination of the data, and restraint before broadly adopting an invasive technology that will add cost and complexity of care, albeit with unproven value, until there is sufficient and convincing evidence for its need and its effectiveness.

References:

1) Hydrogel Spacer Prospective Multicenter Randomized Controlled Pivotal Trial:
Dosimetric and Clinical Effects of Perirectal Spacer Application in Men Undergoing
Prostate Image Guided Intensity Modulated Radiation Therapy. Neil Mariados, et al. IJROBP 2015 in press

2) SAS 9.2 User's Guide, Second Edition (2014)

3) McDonald, J.H. 2014. Handbook of Biological Statistics (3rd ed.). Sparky House Publishing, Baltimore, Maryland (pp. 94-100).

4) Norman GR, Sloan JA, and Wyrwich KW. Interpretation of changes in health-related quality of life: the remarkable universality of half a standard deviation. Med Care, 41 (2003), pp. 582–592.

5) Sanda MG, Dunn RL, Michalski J, et al. Quality of life and satisfaction with outcome among prostate cancer survivors. N Engl J Med, 358 (2008), pp. 1250–1261.

Post Edited (Tall Allen) : 2/14/2016 8:02:00 PM (GMT-7)