Early stage glottic cancer is a highly curable malignancy which can be treated with either larynx sparing surgery (laser excision, cordectomy, or hemilaryngectomy) or radiation . Because there is not a randomized trial to guide treatment decisions, the management of this disease remains controversial. However, at most institutions, radiotherapy is still considered the mainstay of treatment. Because both treatment modalities offer similar rates of cure, decisions regarding which therapy to pursue often lie on the anticipated toxicity profile of a particular regimen. Other factors such as tumor location and extent of disease, co-morbid illnesses and physician and patient preference also impact the final treatment decision. The ultimate goal of any therapy is cure, larynx preservation, high voice quality and overall high quality of life.
Radiation therapy has typically been delivered using a pair of lateral opposed, low energy photon fields that encompass the entire larynx (cobalt to 6 MV) as seen in Figure 3a and 3b. Typical field sizes range from 5 × 5 cm to 6 × 6 cm. Fifteen or 30-degree wedges are often used and improve the dose homogeneity throughout the vocal cords, especially for mid and posterior tumors. The superior and inferior borders are traditionally placed at the top of the thyroid cartilage and bottom of the cricoid cartilage, respectively. Anteriorly, a 1 cm flash with bolus is used. Posteriorly, the field edge is usually placed between the anterior edge of the vertebral body and the middle of the vertebral body. This treatment has consistently produced excellent outcomes with local control rates of 90-95% for T1 lesions and 75-80% for T2 lesions.
Given the excellent results with conventional treatment, some have been reluctant to change technique. In a recent editorial, Feigenberg et al thoughtfully outlined why IMRT offers little benefit and may in fact be of detriment . The authors outline several key arguments in their paper which we will address in our Discussion. First, how can IMRT or any other form of conformal radiation improve upon the excellent rates of local control already achieved with conventional techniques? Second, will the routine use of IMRT lead to a higher risk of marginal failures and lower rates of local control due to smaller planning target volumes? In addition, will IMRT underdose the skin and anterior commissure due to limitations in dose-calculating algorithms, resulting in more local failures? Finally, can IMRT further reduce the risk of major morbidity from the already low rate?
We have shown in this paper that, if a physician is confident in the appropriate PTV to be used for treatment planning, IMRT results in better target coverage than conventional planning. In a tight clinical plan with the posterior border placed at the anterior edge of the vertebral body, PTV coverage is compromised. In this study, the tight clinical plans resulted in a D95 of 60% of prescription and loose clinical plans resulted in a D95 of 86% of prescription to the PTV. In contrast, the IMRT plans were optimized to a D95 of 100%. Also, in the superior-inferior direction, standard field sizes can lead to tumor under-dosing, particularly for bulky T2 lesions with significant supra- or sub-glottic extension. It is well documented that the local control rate after definitive radiation is considerably lower for T2 tumors than T1 tumors. This, in part at least, results from inadequate target coverage for bulkier lesions, particularly since a standard expansion from 5 × 5 cm (T1 tumors) to 6 × 6 cm (T2 tumors) is used with no alteration of the posterior border. This "fixed" increase in field size almost certainly does not adequately account for the differences in the extent of tumor in all cases. Finally, it is evident that regardless of the PTV expansion, the Dmax, and thus the magnitude of the hot spot was less with the IMRT plans. Indeed, when using a clinical wedge plan, the hot spots can approach 12%, which could also compromise long-term vocal function.
Another situation in which IMRT may hold advantages is in patients with thick, short necks. Because of the difficulty with hyperextension, lateral beams cannot cover the inferior aspect of the field due to shoulder obstruction. In these cases, anterior oblique beams are usually used to cover the inferior extent of the target volume. This leads to increased dose to the lung apices. This strategy was indeed utilized on one of the patients in this analysis. In addition to this study, other investigators have examined techniques to maximize the therapeutic ratio in the case of shoulder obstruction. For example, Yom et al. reported outcomes using a "caudal tilt" technique in the postlaryngectomy or pharyngectomy setting. The technique involves the angling of noncoplanar beams in the caudal direction while using 3D planning to deliver dose inferior to the standard three-field match line. The authors reported high 2-year locoregional control rates while shielding a larger amount of posterior lung as compared to the standard 3-field technique . These same principles are used when altering beam angles in the IMRT setting.
With proper target delineation and adequate margins, IMRT should not lead to higher rates of marginal failures and may improve upon the already high rates of local control. The concern about marginal failures was present when IMRT was introduced into routine practice for each tumor site. However, there is no evidence that IMRT leads to higher rates of marginal failures in any disease site [4–6]. On the contrary, IMRT seems to have increased tumor control in both prostate and head and neck tumors by allowing for dose escalation and better target coverage. A number of papers in the past have shown that there is a dose response relationship for larynx cancer, particularly in terms of utilizing a higher dose per fraction [7–9]; thus, proper coverage of the target is critical for definitive radiotherapy in order to maximize local control and minimize patients who will need a laryngectomy.
As a second analysis of this study, we compared IMRT with a 3D conformal technique. Indeed, many of the issues pertaining to conventional techniques, such as dose tradeoff between target structures and normal tissue, and the need for larger margin volumes, would be addressed by the latter technique, in which normal structures could be specified and constraints set to achieve dose escalation. Indeed, while caution should be used and individualized based on the physician's comfort level, CTV to PTV margins as low as 0.3 cm have been utilized with conformal techniques.
We found in this study that, while IMRT and 3D conformal techniques were similar in terms of target coverage and the "clinically meaningful" dose to normal structures (the Dmax to the spinal cord was well below tolerance in all techniques), IMRT demonstrated a significant improvement in terms of the dose to the carotid arteries. For example, a common belief is that, when CT based planning is utilized, an appropriate CTV to PTV margin is 0.5 cm. Even at these relatively tight margins, the mean dose to the carotid arteries was almost 2000 cGy lower when utilizing IMRT than the 3D plan. This dose was lowered even further when an arytenoid sparing plan was utilized, in the case of a patient with a T1N0 lesion located anteriorly.
There is sufficient data that high dose radiation to the carotid arteries can lead to vascular disease. Several reports have shown that head and neck radiation using conventional techniques can cause carotid artery stenosis and increase the risk of ischemic stroke [10–13]. Dorresteijn et al assessed 367 patients treated with radiotherapy for head and neck tumors, including 162 patients with larynx carcinomas, and examined the risk of ischemic stroke. The authors found that the relative risk of developing an ischemic stroke in the patients treated for larynx cancer was 5.1, which reached statistical significance . In a more recent study, Smith et al. examined the risk of a cerebrovascular event in patients older than 65 who previously received head and neck radiotherapy. The authors found that the ten-year incidence of cerebrovascular events was 34% in patients treated with radiotherapy alone, compared to 25% and 26% in patients treated with surgery and radiation and surgery alone, respectively .
Improving clinical toxicity outcomes by decreasing the dose to normal structures has a precedent in head and neck cancer. Most notably, IMRT is routinely used in locally advanced disease to spare the parotid glands and improve salivary function. More recently, investigators from the University of Michigan have also shown that IMRT can decrease the dose to the pharyngeal constrictor muscles, potentially decreasing rates of long-term dysphagia . In the current dosimetric comparison, we show that IMRT markedly reduces the dose to the carotid arteries compared with conventional radiation without compromising coverage of the PTV. Based on this, it is reasonable to postulate that this reduction in dose will decrease the future rate of radiation related carotid artery disease.
Finally, the concern that IMRT will under-dose the skin and anterior commissure is reasonable but the data in this study suggests that with careful treatment planning, this can be avoided. We have shown that IMRT provides at least equal overall coverage of the entire larynx when compared to 2D techniques, as delineated by our PTV, which includes the anterior portion of the structure. Furthermore, as is the case with non-conformal treatment planning, the routine use of bolus provides an additional safeguard to underdosing anteriorly, though due to the unreliability of dose quantification in the buildup region, the extent of dosimetric improvement in this region is not clear.
It is important to note that our study complements the data of a recent study by Rosenthal et al., which demonstrated that intensity-modulated radiation therapy consistently reduces radiation dose to the carotid arteries, with no compromise in tumor coverage. Furthermore, that study demonstrated that radiation planning and treatment times were similar using conventional techniques versus IMRT . Our study expands on the previous one from a planning standpoint by also including a 3D plan comparison and an analysis of the arytenoid dose, and taken together the conclusion of these two studies is that IMRT can spare normal tissues in early stage laryngeal disease without a decrease in tumor dose, both compared to conventional techniques and 3D conformal therapy.
There are several limitations to the current study. First, we compared techniques in only three patients. In order to maximize the generality of our recommendations, we attempted to select patients with normal anatomy, and patients with both T1 and T2 disease. Second, the impact of organ motion was not assessed in this study. Clearly there is some degree of organ motion when treating the larynx, and the typical boundaries with conventional techniques account for this motion. Whether there is a role for on-board imaging with IMRT will be the subject of a future study. Third, we assessed only one 3D conformal beam arrangement, with the purpose being to compare conventional fields with and without normal tissue optimization/CT planning. The addition of more beams to the 3D conformal plans may offer a better dose distribution, though would likely not have the necessary conformality needed to spare the carotid arteries to the extent of IMRT, as demonstrated in Figure 3d.
Finally, we have shown that IMRT provides dosimetric advantages compared with both 2D and 3D conformal techniques, but the clinical significance of such dose reduction is not known. One criticism of our findings may be that it is no surprise that when comparing target and normal tissue dosimetry between two dimensional and conformal techniques (3D and IMRT), the latter methods provide a more optimal dose distribution from a physics standpoint. However, we believe that this study is important because it has shown that when using "appropriate" tumor margins for this disease, IMRT can provide potential long-term clinical advantages even in the context of the relatively small fields and the unique anatomic relationships that are present in the treatment volumes for early-stage glottic laryngeal cancer. We also understand that the utility of IMRT in this disease, and any recommendations that can be drawn from this study, will depend on defining the adequate target volume. Extrapolating from other head and neck sites in which IMRT is utilized and in which excellent rates of local control have been achieved, we believe that CTV to PTV margins of 0.5 - 1.0 cm are reasonable for glottic laryngeal cancer. With this underlying assumption, our data supports the recommendation that IMRT should be strongly considered for this cohort of patients, Based on the dosimetric findings in this study, a reasonable cost-effective treatment paradigm would be: 1) A 2D "tight" clinical plan, a 3D conformal plan with a 0.5 cm margin, or, ideally, an IMRT plan with arytenoid sparing in the case of anteriorly located T1N0 disease, and 2) IMRT in the case of posteriorly located or bulky T2 lesions, where this technique could be used to spare the carotid arteries better than 2D or 3D conformal plans. Perhaps the most important conclusion that can be drawn from this study is that regardless of what is determined to be the appropriate margin in delineating the CTV (and thus the PTV) for early laryngeal cancer, IMRT maximizes the freedom of the clinician to choose a margin that is most appropriate for them. Or put another way, the more confident a clinician is about the PTV, the more of an advantage IMRT offers over other techniques.