CER Update Strength of Evidence Relative to CER No. 20 Findings

Table 11 provides a summary of the conclusions we drew for the relevant interventional comparisons for each Key Question in CER No. 20 and in this update. Because 2DRT and SBRT are not commonly addressed in CER No. 20 and the update, they are not included in Table 11. Moderate strength evidence from the update shows a reduction of the incidence of late grade 2 or higher xerostomia with IMRT compared with 3DCRT. This increases the SOE on this toxicity from CER No. 20, raising it to “high” based on a body of evidence including 2 RCTs and observational studies that are in agreement on this outcome. Evidence in the update is insufficient to show a difference between IMRT and 3DCRT in overall survival or locoregional tumor control rates. We found no new evidence to alter any conclusions of CER No. 20 for any other toxicity, oncologic outcomes, or comparisons.

Table 11

Comparison of relevant CER No. 20 and update conclusions.

Applicability of the Findings

In general, applicability assessment would depend on a body of evidence sufficient to form new conclusions about the comparative outcomes of 3DCRT, IMRT, SBRT, and PBT in treatment of head and neck cancer. However, comparative evidence that meets study selection criteria for this CER update is sparse for 3DCRT, IMRT, and SBRT, and nonexistent for PBT. In the absence of sufficient evidence, additional factors may be considered in making a treatment decision. Those could include relative convenience and cost, issues outside the scope of this CER.

In preparing this update, we reconsidered the RT modalities included in CER No. 20 and whether all remained applicable to current radiation oncology practice. In particular, we examined the role of opposed beam 2DRT in modern radiation oncology practice. Based on the current literature and input from our TEP members, we concluded 2DRT is no longer used in the U.S. for definitive treatment of head and neck cancer, thus we excluded it from the update. We realize in doing so we excluded evidence from a RCT performed in China that showed a statistically significant improvement in overall survival with IMRT compared to 2DRT, which to our knowledge is the only study that has shown a statistically significant survival benefit of one RT modality compared to another.⁵⁴ However, this did not alter our overall conclusion to exclude 2DRT from the current report.

We also re-examined whether to include brachytherapy in this update; it was not part of CER No. 20. Although brachytherapy can be used in select cases as a means of dose escalation in conjunction with external beam irradiation for head and neck cancer,^3,4 this practice has become uncommon because sufficient dose escalation can often be achieved in these cases with a noninvasive approach (e.g., conformal RT). Brachytherapy alone is very rarely employed, except in small (T1) tumors of the nasal vestibule, lip, or oral cavity, which are relatively uncommon (1 percent to perhaps 5 percent of all cases).^5–9 Therefore, because use of brachytherapy alone for primary management of head and neck malignancies has limited applicability in modern head and neck radiation oncology practice, we did not seek evidence of it for this CER; we focused instead on RT modalities that are used as the sole RT modality for a given presentation of head and neck cancer.

We considered including dosimetry studies in CER No. 20, and this update. For both reports, our ultimate conclusion not to include dosimetry studies was agreed upon among our EPC team, among AHRQ personnel, and in discussion with our TEP. The primary rationale for this conclusion is that dosimetry studies do not provide a link to actual clinical outcomes that are realized by patients. Dosimetry modeling is clearly needed to advance research in RT methods, but it does not provide evidence for clinical efficacy.

Key Questions 1 and 2

The degree to which the evidence presented in this update is applicable to clinical practice is a function of the similarity between populations in the included studies and the patient population that receives clinical care in diverse settings. It also is related to the relative availability of the modalities. Because of the overall weakness of evidence for Key Questions 1 and 2, we have primarily limited comments to the relevance of the PICOTS elements, a practical and useful structure to review the applicability in a systematic manner (Table 12).

Table 12

Summary of applicability of evidence for Key Questions 1 and 2.

Findings in Relationship to What Is Already Known

Our updated systematic literature search and review revealed no relevant evidence-based guidelines we could compare with our findings for any of the Key Questions.

Limitations of Current Review and Evidence Base

The primary limitation for all Key Questions here is lack of well-designed and conducted comparative trials. Although the body of evidence we identified was more substantial for 3DCRT and IMRT than SBRT, and nonexistent for PBT, we have significant concerns about interstudy heterogeneity, with variability in RT dose, schedule of treatment, concurrent treatments, patient selection criteria, tumor size and location, and so forth.

We acknowledge that our inclusion of comparative studies alone may have limited collection of RT-associated adverse events that may be revealed in larger observational studies. However, we believe our decision to focus on key comparative outcomes xerostomia, dysphagia, and salivary gland toxicity was merited based on our understanding of the literature and the importance of those toxicities to cancer patients.

As stated previously in this report, we are not sure whether the inconsistency we observed in comparative RT-associated adverse events reflects a lack of systematic collection of this type of information by investigators, or failure to consistently report it in publications. In a systematic review in general, a heterogeneous evidence base makes it very difficult to assess the relative benefits and harms of any modality, particularly evidence drawn from nonrandomized trials, and to assess the SOE of a body of evidence. In this CER update, the sparse new evidence we identified limited additional comparative assessment among the modalities. We therefore believe further careful study of the RT methods compared in this CER is needed, particularly in the regimens of Key Question 1 or 2 to establish optimal technical protocols and patient selection criteria, perhaps standardizing and comparing them across institutions. These data and methods could, in theory, be applied to the design and conduct of comparative studies, as outlined in the Research Gaps section below.

We are aware that a body of dosimetry evidence is available to suggest potential differences in the benefits and harms of different conformal RT types. Our exclusion of such evidence may be viewed by some readers as a limitation of this CER update. However, we maintain that because dosimetry modeling studies do not provide a clear link to clinical outcomes, they do not add critical information to assess the comparative effectiveness of RT in the treatment of head and neck cancer.

Research Gaps

The primary research gap we identified is a continuing lack of evidence from well-executed comparative studies (randomized or otherwise) to draw conclusions on the relative clinical benefits and harms of the RT methods used in patients with head and neck cancer. We also identified some feasibility issues associated with the RT methods that are potential impediments to the type of rigorous comparative studies we suggest are necessary to determine their comparative effectiveness. In this section, we first describe characteristics of ideal comparative studies we believe are needed to compare these technologies. Some potential impediments to such studies are discussed subsequently in this section.

Summary and Conclusions

Key Questions in CER No. 20 asked whether any of the RT modalities under consideration (2DRT, 3DCRT, IMRT, PBT) is more effective than the others:

• in reducing normal tissue toxicity and adverse events, and improving QOL
• in improving local tumor control, time to disease progression, and survival
• when used in certain anatomic locations or patient subpopulations
• whether there is more variation in patient outcomes with any modality secondary to user experience, treatment planning, or target volumes

The main finding of CER No. 20 was that late grade 2 or higher xerostomia was reduced and QOL domains related to xerostomia were improved in patients treated with IMRT compared with those who received either 3DCRT or 2DRT. Evidence was insufficient to draw relative conclusions on survival or tumor control; adverse events other than late xerostomia (e.g., mucositis, dysphagia, skin toxicities, osteoradionecrosis of the jaw); whether patient and tumor characteristics affected relative outcomes; or whether physician experience and treatment characteristics affected relative clinical outcomes such as survival or treatment-associated adverse events.

Moderate strength evidence from the update shows a reduction of the incidence of late grade 2 or higher xerostomia with IMRT compared with 3DCRT, which strengthens the conclusion on this toxicity from CER No. 20. Evidence in the update is insufficient to show a difference between IMRT and 3DCRT in overall survival or locoregional tumor control rates. We found no new evidence to alter any conclusions of CER No. 20 for any other toxicity, oncologic outcomes, or comparisons.