Late radiation morbidity following randomization to preoperative versus postoperative radiotherapy in extremity soft tissue sarcoma
Reviewer: John Wilson, MD Abramson Cancer Center of the University of Pennsylvania
Authors: Davis AM, O'Sullivan B, et al. Source: Radiother Oncology, 2005 Apr;75(1):48-53.
Initial paper review
This is the update of the SR.2 trial sponsored by the NCI and headed by Dr. Brian O'Sullivan from the Princess Margaret Hospital (PMH) in Toronto.
It compares preoperative (preop) versus postoperative (postop) radiotherapy (RT) for soft tissue sarcomas of the extremities only.
An earlier report of this study, published in Lancet in 2002, involved a total of 190 patients (pts). Pts were stratified by tumor size (smaller or larger than 10 cm), and then randomized to either preop or postop RT.
Pts were enrolled between 1994 and 1997.
The preop group was treated with 50 Gy in 2 Gy fractions, with proximal and distal margins of 5 cm each. Preop pts with positive margins at surgery also received a boost of 16-20 Gy, with 2 cm margins.
In contrast, all postop pts received both the initial 50 Gy as well as the boost of 16-20 Gy.
The surgery and RT were done 3-6 weeks apart, and preop pts who required a radiation boost were only treated after their wounds had healed.
A longitudinal strip of skin and subcutaneous tissue (of unspecified size) was left untreated for at least half of the course, unless the margin around the target region was less than 2 cm at any point that was not confined by an intact fascial boundary.
Interestingly, participating surgeons had to agree to disregard the randomization when planning the resection and wound closure.
The primary endpoint was presence or absence of a major wound complication, defined as:
a secondary operation under general or regional anesthesia for wound repair (debridement, operative drainage, and secondary wound closure)
invasive procedures, i.e. aspiration of seromas
readmission for wound care, i.e. IV antibiotics
persistent deep packing for 120 days or longer
These wound complications were assessed only up to 4 months after surgery in the earlier report.
Secondary endpoints included:
The groups appeared well-balanced for demographic characteristics.
After a median f/u of 3.3 yrs, there were:
Significantly more wound complications in the preop group (35 vs 17%, p=0.01).
However, acute toxic skin effects (grade 2 or greater) were more common in the postop group, (68 vs 36%, p<0.0001) likely related to the higher dose and larger volume of treatment.
There was no difference seen with respect to local recurrence, regional or distant failure, and progression-free survival.
However, interestingly, the overall survival was just barely significantly better in the preop group (p=0.048), but no actual percentages were given. This may have been a statistical anomaly, as the tumor control didn't seem improved at all.
The update was published three years after the preliminary report.
This paper set out to test the hypothesis that at 2 yrs out, patients treated with postop RT would have more chronic radiation morbidity such as:
...And they would thus have poorer extremity function.
Of the 190 pts enrolled, 181 were treated in accordance with protocol, and 161 pts were assessed at 2-year follow-up from the completion of the first 50 Gy of radiation.
There were no significant differences in patient characteristics, but there were fewer patients assessed in the postop arm (56 vs. 73) for reasons such as death and being lost to follow-up.
Late subcutaneous fibrosis and joint stiffness were assessed using the joint EORTC/RTOG late toxicity scoring criteria.
Fibrosis ranges from subcutaneous fat loss to field contracture of > 10%. on linear measurement.
Joint stiffness was more subjective, ranging from slight decrease in range of motion loss without pain, to pain and "severe stiffness".
Edema was measured according to Stern's criteria
Mostly subjective, but with grade 4 it was more defined, with shiny and tight skin or cracking of the skin.
A priori, it was determined that grade 2 or greater toxicity was clinically significant.
Patient function, physical disability, and general health status were evaluated using two scales: the Musculoskeletal Tumor Society Rating Scale, or MSTS, and the Toronto Extremity Salvage Score, or TESS.
The MSTS is a clinician-rated measure of impairment at the organ level, ranging from 0-35.
The TESS is a patient-rated measure and ranges from 0-100.
For both scales, higher scores implied better function.
There was no statistically significant difference between treatment groups for anything, but there was a trend to worse subcutaneous fibrosis with postop RT (48.2 vs. 31.5%, p=0.07), and every other endpoint also showed a slightly better, though non-significant, difference in favor of preop radiation.
Patients who had grade 2 or greater toxicities were more likely to have worse functional scores with both scales, but there was no significant difference in the scales by treatment arm.
On logistic regression, only field size was a significant risk factor for subcutaneous fibrosis and joint stiffness; Dmax was not significant (actually called D T,max, which is the maximum dose to at least 2 cm 2 on the three cross- sectional isodose distributions).
The authors report that these results "suggest" that postop RT causes more late toxicity, but they admit the toxicity is not significantly different from that seen with preop RT.
However, they do point out that the sample size was fairly underpowered to show a difference in late toxicity, with powers of 40% for fibrosis, 12% for joint stiffness, and 22% for extremity edema.
They felt that the toxicities they saw were similar to those reported by others.
Although the Dmax was not significant in predicting complications, they comment on another study (Stinson) that showed a correlation between higher dose and pain, decreased strength, and decreased range of motion.
They conclude by saying that preop RT may have less late toxicity, but this must be balanced against a higher risk of wound complications.
They also say that there might be a life-long risk for these late complications, and that longer follow up is needed to assess this probability.
One of the problems with this study was that there was no mention of enforcing blind physician assessments. In light of this, combined with the mostly subjective scales that were used, there is a risk of significant bias when evaluating late toxicity. Given the authors' hypothesis that late toxicity is worse with postop, then, if anything, they would be more likely to score postop toxicity more harshly.
One could argue against the relevance of the physician bias given that both the patient and physician scales were slightly (but nonsignificantly) in favor of preop RT.
Having said that, the argument remains as patients may also have a bias in favor of preop, just like the physicians.
Currently, O'Sullivan is conducting a study at PMH in which patients assigned to get preop treatment are partially planned ahead of time by the surgeon, such that the area of tissue that will eventually form the flap receives a lower dose of radiation (but only if it is far enough from areas of concern for tumor involvement).
A review of this article in Cogent Medicine comments that this was essentially an underpowered study, and the late toxicity probably is significantly worse with postop treatment.
"Although the initial report from this study showed an advantage to postoperative therapy in terms of wound complications, the decrease in late effects with preoperative therapy is, in my view, more compelling." He continues by saying "I maintain my preference for preoperative therapy..." – Dr. Joel Tepper