Inherently, MLSs are very chemotherapy-responsive tumors, and this modern series confirms that finding. The dose-intensive regimen of doxorubicin (75-90 mg/m2) and ifosfamide (10 gm/m2) in this study attained a PR rate of 43.2% by RECIST and 86.5% by Choi criteria for the entire cohort. No patient had progressive disease as the initial response to therapy. The efficacy of this regimen was not statistically different between patients with resectable disease and patients with advanced disease. We were encouraged by the high response rate, accompanied by a lack of disease progression, observed during the first four to six cycles, which highlights the utility of chemotherapy in the preoperative setting, in which chemotherapy is likely to shrink the tumor or kill tumor cells.
The high efficacy of dose-intensive AI in MLS appears to be not a feature unique to this regimen but rather a feature of the chemotherapy sensitivity of MLS. Patel et al.  reported a response rate of 44% (by RECIST) in 18 patients treated with first-line doxorubicin (60-90 mg/m2 as a 48- to 96-hour continuous infusion) and dacarbazine (900-1000 mg/m2) with or without cyclophosphamide (600-750 mg/m2). Grosso et al.  reported a high response rate of 51% per RECIST in a retrospective study of patients treated with trabectedin and noted that a substantial number of these patients (74%) demonstrated what their group termed tissue response prior to tumor shrinkage. These tissue responses would all qualify as Choi responses. Similar response rates were recently reported in a prospective study by Gronchi et al. . In that phase II study, 13 patients were treated preoperatively with first-line trabectedin (1.5 mg/m2 every 3 weeks) and had a 46% response rate by RECIST with no progressions noticed during the neoadjuvant period.
In addition to being chemotherapy sensitive, MLSs are also quite radiotherapy sensitive, as demonstrated by Guadagnolo et al. . In their study, radiotherapy given before or after surgery to 128 MLS patients resulted in a 97% 5-year local control rate, accompanied by an 81% 5-year disease-free survival rate and 87% overall survival rate. The high efficacy of chemotherapy and radiotherapy in MLS suggests they should play a role whenever surgical morbidity or organ functionality preservation issues arise in the neoadjuvant setting.
Based on the definitions of the Choi criteria, it is not a surprise that their use results in a significantly higher response rate than is seen with RECIST. Choi criteria set the bar for PR lower, identifying PR as a 10% decrease in tumor size, compared with 30% using RECIST. Bearing this difference in mind from a clinical perspective, Choi criteria better mirror day-to-day practice in the clinic, where decisions about chemotherapy continuation often need to be made based on subtle changes in tumor size. However, the major caveats for the Choi criteria seem to be twofold: First, although the method has been validated in assessing responses in gastrointestinal stromal tumors treated with imatinib mesylate , it is not yet widely accepted nor utilized for monitoring responses to chemotherapy in other tumor types. However, we believe that monitoring and reporting responses by both RECIST and Choi criteria, as was done in this study, is more informative than RECIST alone because the Choi criteria capture more subtle changes in tumor response and provide a more accurate assessment for oncologists and their patients regarding the probability of benefit from chemotherapy.
The second caveat for the Choi criteria is that for determining density, a CT scan with contrast is required. This aspect of the Choi criteria is less amenable to primary MLS since the majority of tumors originate in an extremity, where MRI is often the preferred imaging modality. Nevertheless, since changes in tumor size may be insufficient to represent actual tumor response, Stacchiotti et al.  suggested MRI with contrast material enhancement to complement tumor size, thus making Choi criteria a predictive tool of pathologic response in high-grade STS. Interestingly, decreased contrast enhancement, either on CT or MRI, following chemotherapy appears to be a predictor of pathological response.
The current study was limited by its retrospective nature that precluded strict follow-up as well as toxicity monitoring. Nevertheless, from our long-standing experience with this dose-intensive regimen, its toxicity is manageable and involves mostly myelosuppression; only two patients out of 37 in our study had to discontinue treatment, and in both cases discontinuation was secondary to prolonged myelosuppression. Severe potential forms of toxicity related to this regimen, such as cardiotoxicity, nephrotoxicity, and neurotoxocity, are relatively rare in our sarcoma center due to stringent follow-up and the employment of a series of preventive measures, such as the use of continuous-infusion doxorubicin treatment over 72-96 hours or bolus administration with dexrazoxane (Zinecard) in lieu of bolus doxorubicin alone; generous hydration combined with forced diuresis when needed; monitoring of fluid status and creatinine levels on a daily basis; and supplementing with bicarbonate and electrolytes such as potassium, magnesium, and phosphorus. Finally, setting a serum albumin lower limit of 3 gm/dl for initiating treatment with ifosfamide and supplementing with intravenous albumin significantly decrease neurotoxicity. In addition to the regimen-related preventive measures above, routine supportive care, including antibiotics, antiemetics, and growth factors, is also provided to all patients.
This study was also limited by the small number of patients, which precluded further analysis for any association between response and other variables such as age, sex, round-cell change/hypercellularity, doxorubicin dose (75 or 90 mg/m2), or number of cycles.
To our knowledge, this is the first report describing response to first-line AI in MLS. By reporting responses by both RECIST and Choi criteria, we were able to compare our findings with other studies that used the more conventional RECIST alone. We conclude that MLS is a chemotherapy-responsive tumor and that AI dose-intensive chemotherapy is an effective and safe approach for MLS. We advocate the use of more sensitive criteria such as the Choi criteria in evaluating response to chemotherapy.