Comprehensive Head and Neck Radiotherapy Dose-Volume Constraints Do Not Apply to Smaller Volumes

Discussion

Studies correlating RT doses to different organs with long-term vasculopathy and swallowing function outcomes in patients treated for head and neck cancer patients have resulted in conflicting data. Structures implicated in previous studies mainly include the pharyngeal constrictor muscles (superior, middle and inferior), as well as the larynx. However, other structures may play important roles in swallowing. Due to the use of concurrent chemotherapy, altered fractionation and escalated RT, the rates of long-term dysphagia have significantly increased (1). Generally, dosage exceeding tolerance of the constrictor muscles and larynx is thought to be a major etiology for long-term dysphagia. Multiple studies have shown that patients with locoregionally advanced head and neck cancer treated with definitive RT, with or without chemotherapy or biological agents, demonstrate late severe dysphagia in up to 40% of cases. Authors have advocated that both laryngeal and pharyngeal constrictor doses are responsible and predictive for the development of late severe dysphagia (2-6). There is sufficient data that high-dose RT to the neck and carotid arteries can lead to RICV. These studies have shown that RT to the neck increases the risk of RICV to a statistically significant level but to a small magnitude (7-12). This retrospective review included patients with early glottic SCC for whom the general institutional policy was to treat with definitive RT with once-daily fractionation. Dosimetric parameters of the main anatomic structures (bilateral carotid arteries, supraglottic, glottic, subglottic, middle and inferior PCs) involved in the RT fields were analyzed to assess RID and RICV.

RID. Our data show that RT doses to the whole larynx (supraglottic, glottic, and subglottic), and PCs (middle and inferior) do not result in long-term dysphagia. Of note, the linear-quadratic model was used to determine the biological equivalent dose (BED) for RT using BED= nd [1+ (d/α/β)]. It was found that 67.5 Gy at 2.25 × 30 is BED to 76 Gy at 2 Gy × 37 and 84 Gy at 1.8 Gy × 47. This calculation was based on an α/β ratio of 3 and 10 for late and early tissue respectively (13). This finding suggests that RT dose cut-off values should be re-assessed for an association with the likelihood of long-term dysphagia. This may be due to the small tumor volume, consequently small RT field, the absence of chemotherapy and its chemosensitization effect. None of the patients in the current study initially had any swallowing impairment before RT and they continued to do very well despite RT with a high dose per fraction to the larynx and PCs.

Pauloski et al. reported that patients with laryngeal and pharyngeal tumors complaining of dysphagia were significantly more likely to have aspiration (up to 30%) versus those without complaints (≤5%), suggesting that this may be a reasonable approximation of the true incidence of aspiration (14).

Other institutional studies examining the association of doses to normal structures and risk of dysphagia found that up to 47% of patients had developed aspiration (2-6, 15, 16). Clinical factors associated with aspiration included primary site, advanced T-stage, larynx and inferior pharyngeal constrictor doses. The studies have shown that a mean dose up to 48 Gy to the larynx results in no aspiration issues (3-6). Using a different laryngeal anatomical definition, the Danish group found that V60 and V65 were associated with aspiration (15). Larynx V55 through V70 was also associated with aspiration (15), whereas in both the Michigan and Harvard experience, larynx V50 was found to be most significant (3, 4).

The threshold percentage for larynx V60 and V65 of 77% and 71% respectively in our study was much higher than the larynx V50 threshold of 21%, 26%, and 32%, reported by other investigators (3-6). Dissimilar to Caglar et al. (4) and Caudell et al. (5) we found that the dose to the PCs was not a predictor of dysphagia. Whereas they found V50, V60, and V65 to be significant, we found volumes irradiated to higher (V50, 60 and 65) doses not to be associated with any long-term dysphagia (4, 5). Table III summarizes the most important studies on RID. The difference in results may be the result of variances in patient populations, treatment modality, and structure delineation.

RICV. Smith and Loewenthal, in 1950, were the first to report on the adverse impact of RT on elastic arteries in irradiated mice (17). It was later reported by other groups that RT induced atheromatous lesions in the arterial wall of irradiated vessels (18-20). Clinically, this was first reported by Glick, who reported a case of bilateral carotid occlusive disease following RT for vocal cord cancer (21). This was followed by other reports that confirmed the existence of RICV (22-27). Recent reports have shown that RT to the head and neck may cause carotid artery stenosis and increase the risk of ischemic stroke (7-12). The European group, Netherlands Cancer Institute, performed a retrospective study on 367 patients who underwent definitive RT with and without chemotherapy for head and neck tumors, of whom 162 patients had cancer of the larynx (8). Dorresteijn and colleagues reported that 14 ischemic strokes were identified and RT had a statistically significant relationship with ischemic stroke in patients with cancer larynx, with a 15-year cumulative risk of stroke of 12% (8).

More recently, the MD Anderson group performed a Surveillance, Epidemiology, and End Results (SEER)–Medicare cohort study by Smith et al. (11). These authors investigated the risk of a cerebrovascular event in patients older than 65 years who had previously received head and neck RT. The authors found that the 10-year incidence of cerebrovascular events was 34% in patients treated with RT-alone, compared to 25% and 26% in patients treated with surgery-alone and surgery-plus-adjuvant RT, respectively. However, there was no significant increase in cerebrovascular events in patients treated with surgery with or without adjuvant RT (11). Another (SEER database and Medicare claims records) study by the Mount Sinai group reported a statistically significant increase in stroke incidence in head and neck cancer patients treated with RT, but there was no significant difference in stroke mortality: the 10-year rate of stroke was 10% with RT vs. 7.5% without RT (p=0.01) (12). Table IV summarizes the recent RICV studies.

In summary, controversial and contradictory data continue to exist as the results reported herein suggest that the doses to the carotid artery, larynx and PC (middle and inferior) muscles were of limited significance in development of sequlae. It is possible that differences in the patient population, treatment modality and techniques, as well as delineation of various structures, could change the emphasis of various analyses with respect to the significance of dose constraints as noted previously. Our study contradicts those previously published because our patients did not develop problems at the doses that prior studies reported to be problematic. We surmise that the reason for this difference is that complication risk is sensitive to treatment volume in a way that is not taken into account by the typical DVH analyses. We treated only the larynx and our patients did not develop the problems reported when regional nodes and larger volumes of the larynx are treated. This finding is important because it means that the DVH thresholds for clinical complications are only applicable to the specific situations in which the data were generated. Changing the treatment volume will likely change the dose thresholds for complications and the rate of complications at a given dose. Therefore, reducing the treatment volume is likely to be an important strategy for reducing treatment toxicity, independent of other factors.

Of note, all the previous and current DVH values are based on static (initial CT simulation and planning) dosimetry. Given the fact that there are daily variations in patient setup, weight loss, inter- and intrafraction organ motion mainly during swallowing, and tumor shrinkage, we could not ascertain how much dose was deposited to the critical structures reported in our study. This problem is currently being resolved as we have implemented the usage of daily image-guided radiation therapy. This has enabled daily contouring of the structures of interest via on-board imaging, cone-beam CT, to calculate the actual delivered doses for the critical structures. We hope this strategy will be beneficial in determining normal tissue tolerances.

Lastly, the strengths of our study include single-institution experience, a sufficient number of patients, and management of treatment by two radiation oncologists (LBH and KSH). In addition, all contours were drawn by one radiation oncologist (WFM), and an independent statistician (RAS) who was not involved with the data abstraction performed all the statistical analyses.

The caveats of the study include it is a non-randomized retrospective analysis and vulnerability to selection bias. An additional factor is the dependence on accurate documentation. All previously reported studies have shown DVHs from comprehensive head and neck RT with and without chemotherapy or biological agents. However, in the current study, all patients were treated with small-field larynx-only RT. Moreover, there was no pre-RT carotid study to compare to post-RT changes, if any.

RICV was evaluated primarily on a clinical basis and from medical chart review. Furthermore, only 4% of patients agreed to be evaluated by MBS due to lack of symptoms. Four percent is certainly a very small number to draw any conclusion from but in the absence of symptoms, we could not justify the extra cost and radiation exposure due to MBS for all the 253 patients in our cohort. We selected the patients (4%) who received the highest dose to the larynx and constrictors. Lastly, the current study has a relatively short follow-up (median of seven years), which may not be sufficient for development of RICV.