• © 2001 by American Society of Clinical Oncology

Balancing the Possible Effectiveness of Postoperative Radiotherapy for Non–Small-Cell Lung Cancer Against the Possible Detriment of Radiation-Induced Toxicity

THE POSSIBLE role of thoracic radiotherapy as a cause of intercurrent disease and death for patients with non–small-cell lung cancer (NSCLC) has been difficult to assess because these patients have a high incidence of early death owing to lung cancer. These lung cancer deaths often occur before the general time period when one would expect to detect the manifestations of late radiation-induced toxicity. This problem is particularly true for patients with locally advanced and unresectable NSCLC. The risk of early death from lung cancer is not quite as daunting for patients who have undergone a complete resection of NSCLC, especially in the setting where lymph node metastases are absent. Therefore, a significant proportion of patients in the latter group may have sufficiently long survival time to allow for adequate assessment of treatment-related late toxicity. In fact, the suggestion that high-dose (approximately 60 Gy) postoperative radiotherapy (PORT) may be detrimental to node-negative patients was made more than 20 years ago.¹ Further assessments of treatment-related toxicity in this node-negative population have been limited by the fact that most groups abandoned the use of PORT for these patients in the early 1980s.

Because PORT is not standard practice for node-negative patients, recent investigations of PORT-related toxicity have involved patients with completely resected node-positive disease. Even though these patients do have a fairly high rate of early mortality from lung cancer, attempts have been made to distinguish treatment-related deaths from deaths owing to lung cancer. Also, investigations have been performed to determine whether certain treatment-related factors, such as PORT dose and fractionation, are associated with treatment-related deaths.^2,3 A better understanding of these factors may help physicians tailor radiation treatments for individual patients based on their specific potential risks and benefits. The article of Machtay et al⁴ adds to our understanding of the potential factors that may be associated with PORT-induced deaths for patients who have undergone a complete resection of NSCLC. If it is assumed that the authors used fairly consistent target volumes during the study period, the assessment of dose effects is an important attribute of this analysis. They have suggested that PORT doses of ≥ 54 Gy may be potentially more detrimental with respect to intercurrent disease when compared with lower doses. Although they were not able to make firm conclusions about the cause of the intercurrent deaths that were associated with the higher doses of PORT, others have suggested that cardiorespiratory deaths are important events for these patients.³

The authors attempted to quantify the risk of intercurrent disease in the population of patients who received PORT compared with United States vital statistics. Their analysis included a correction for the gross differences in smoking behaviors between the two populations. Additionally, they made a modification of the United States vital statistics to delete the lung cancer deaths by using a correction factor that was uniformly employed across the population. Finally, they reported that the risk of intercurrent deaths in the PORT population was not significantly increased compared with the corrected United States vital statistics. They did find that there was a 3.5% increase in intercurrent deaths in the PORT group (not statistically significant). This increase would have represented a statistically significant increase if this same rate had been found in a larger study population comprising three times as many patients. The authors acknowledged that this comparison was fraught with potential problems, as the PORT group and the general United States population were inherently different populations, even after corrections were made for smoking behavior and lung cancer risks. However, the comparison may allay some fears that PORT greatly increases intercurrent deaths.

Considering the fact that thoracic radiotherapy is commonly used in the United States for both locally advanced NSCLC and in the PORT setting (see survey results below), the analysis of the relationship between PORT doses with respect to the incidence of intercurrent disease may be the greatest contribution of the article by Machtay et al.⁴ The dose-response data suggested that lower doses of PORT (< 54 Gy) may be associated with substantially lower risk of intercurrent deaths (0%, 4-year actuarial rate) compared with higher doses of ≥ 54 Gy (17%, 4-year actuarial rate) (P = .06). The calculated expected actuarial rate of intercurrent deaths was 10% at 4 years based on United States vital statistics.

This dose-response information gives further strength to the contention that several large randomized trials have shown associations between PORT and increased intercurrent disease, because these trials used high doses of PORT.⁵ These same trials were the subject of a meta-analysis, which concluded, “Postoperative radiotherapy is detrimental to patients with early-stage completely resected NSCLC. . . . the role of PORT in the treatment of N2 tumors is not clear. . . ”⁶ The work of Machtay et al⁴ raises the possibility that the meta-analysis found PORT to be detrimental, in some groups of patients, solely because of the PORT technique. This meta-analysis included nine randomized trials in which PORT was compared with observation. The investigators calculated hazards ratios to determine the effect of treatment on survival for the individual trials as well as the overall group. Examination of these hazards ratios revealed that the five trials with the greatest hazards ratios (associated with PORT-related survival decrement) included PORT treatments of ≥ 54 Gy, and some included 60 Gy of PORT with individual daily fractions of 2.5 Gy (conventional daily fractions in the United States are often 2 Gy). Therefore, the use of high doses of PORT to critical normal structures may be the greatest factor involved in PORT-related intercurrent deaths. These intercurrent deaths may prevent the possibility of detecting a potential benefit of PORT-related increased local control.

The hypothesis raised by the work of Machtay et al⁴ is that PORT doses of less than 54 Gy are safe, with minimal or no increased risk of intercurrent death. This hypothesis is supported by the large series of patients from the Mayo Clinic in which 50.4 Gy was the median dose of PORT.^7,8 The Mayo Clinic experience not only indicated a lack of PORT-associated risk of intercurrent disease, but also demonstrated a substantial survival benefit for patients who received PORT after complete resection of tumors with N2 metastases as compared with patients who were observed after surgery.

The meta-analysis was clearly an important article, which was even reported by many of the lay press organizations in the United States and abroad. Many of the trials that were included in the meta-analysis used high total and daily doses. Also, many of the trials included node-negative patients. Therefore, it was difficult for radiation oncologists who routinely used doses less than 54 Gy, for patients with nodal metastases, to determine the appropriate interpretation of the meta-analysis for their individual practices. Several months after the publication of the meta-analysis in the Lancet, the Department of Radiation Oncology at the University of Alabama at Birmingham reviewed this article during their quarterly State Journal club. After this review, it was decided to survey the Alabama and Georgia Societies of Radiation Oncologists to assess general practices regarding PORT.

A questionnaire was submitted to 134 members, and 45 were returned. This low return rate hindered definitive conclusions regarding standard practices. However, there were certain trends that were interesting. Physicians were queried regarding their opinions about the appropriate use of PORT for various stages of resected NSCLC. The returned questionnaires demonstrated that most of the 45 radiation oncologists agreed or strongly agreed with the recommendation for PORT for patients with completely resected NSCLC with N2 metastases (96%). Additionally, a majority of radiation oncologists agreed or strongly agreed with the recommendation for PORT for patients with N1 metastases (73%). It was determined that the recommendation for PORT occurred with a statistically significant higher frequency for patients with N2 metastases compared with the N1 setting (P < .01, Sign test).⁹ Also, most radiation oncologists did not recommend PORT for the N0 setting. The results of the Alabama/Georgia survey were corroborated by another survey group, which included physicians from the following specialties: medical oncology (42%), gynecology oncology (15%), hematology (10%), radiation oncology (9%), and other specialties (24%).¹⁰ In this latter survey, 73% of physicians recommended PORT for patients with completely resected stage III NSCLC. The appropriate use of PORT remains controversial. However, it seems that a majority of physicians in the United States recommend PORT for patients with completely resected NSCLC with N2 metastasis, although the literature shows significant conflict regarding this therapeutic intervention. An update of the Alabama/Georgia survey is currently being planned.

Therefore, the contribution by Machtay et al⁴ and the former contributions by the Mayo Clinic group^7,8 have suggested that PORT can be delivered safely and may greatly enhance survival in certain populations of patients with resected NSCLC. However, a randomized trial of PORT (< 54 Gy) versus observation is warranted in patients with completely resected NSCLC with N2 lymph node metastases and possibly selected groups of patients with N1 metastases.¹¹ A trial of this nature was recently attempted by the Cancer and Leukemia Group B (CALGB). This trial (CALGB 9734) randomized patients with completely resected N2-positive lymph node metastases to four cycles of paclitaxel and carboplatin versus the same chemotherapy followed by PORT (50 Gy). This trial closed in spring 2001 because of poor accrual. Previously, the North Central Cancer Treatment Group (NCCTG) attempted to assess the efficacy of postoperative chemotherapy (etoposide/cisplatin) or PORT or PORT with chemotherapy in comparison with observation for patients with completely resected node-positive NSCLC (NCCTG, 88-24-53). This four-arm trial did not accrue any patients in 6 months and was subsequently redesigned to a two-arm trial of PORT versus PORT with chemotherapy.¹² Like the CALGB trial, this trial was closed because of poor accrual. The inability of the physicians in the United States to address the efficacy of PORT in a prospective manner has been disappointing. Therefore, until we have solid randomized trials regarding PORT with doses less than 54 Gy, physicians must weigh the risks and benefits of this treatment, realizing that there are conflicting opinions in the literature.