Meta-analysis of Impaired Vocal Cord Mobility as a Prognostic Factor in T2 Glottic Carcinoma

Objectives  To pool and meta-analyze the reported outcomes in patients receiving radiotherapy for T2 glottic carcinoma with impaired vocal cord mobility.

Data Sources  A MEDLINE search and manual search were conducted to identify all studies published between January 1, 1950 and September 30, 2007, in English on the treatment of T2 glottic carcinoma. Search terms included laryngeal neoplasm, glottis, and glottic carcinoma.

Study Selection  All studies of primary radiotherapy for T2 glottic carcinoma with explicit reporting of outcomes with regard to the presence or absence of vocal cord immobility were included in the meta-analysis. Studies with data reported elsewhere were excluded.

Data Extraction  A meta-analysis using a fixed-effects model was conducted for outcome measures of local disease control, ultimate survival after salvage surgery, absolute survival, and disease-specific survival. Odds ratios (ORs), 95% confidence intervals (CIs), and tests for heterogeneity were reported.

Data Synthesis  Thirty-five studies were identified, of which 21 met criteria for meta-analysis. All studies were retrospective. Comparison of 5-year local control of disease for lesions with impaired vocal cord mobility (T2b) vs those with normal vocal cord mobility (T2a) showed a statistically significant difference (OR, 1.83; 95% CI, 1.52-2.20; P < .001). Ultimate control after salvage surgery was reported in 7 studies, which also showed significantly better outcomes for T2a lesions over T2b lesions (OR, 1.90; 95% CI, 1.23-2.92; P = .005).

Conclusions  This meta-analysis indicates that impaired vocal cord mobility has a negative impact on local disease control and ultimate disease control in patients receiving primary radiotherapy for the treatment of T2 glottic carcinoma. We recommend that the subdivision of glottic cancer to T2a and T2b be reviewed and considered by the American Joint Committee on Cancer for inclusion in the Cancer Staging Manual.

In the United States, laryngeal carcinoma is the most prevalent malignant neoplasm of the head and neck region, with approximately 60% of tumors arising in the glottis. Treatment of glottic carcinoma has traditionally used both surgery and radiotherapy, with generally good rates of disease control for early-stage disease.^1-21 Single-modality therapy is typically preferred for early-stage glottic carcinoma.^2,4,5,7,8 Because of the crucial role the larynx plays in swallowing and communication, treatment modalities that allow for at least partial preservation of laryngeal function have now become widely used.^11,22-25

Staging of laryngeal carcinoma has been defined and periodically revised by the American Joint Committee on Cancer and the International Union Against Cancer.²⁶ In the present edition of these systems, T2 glottic carcinoma has been defined as carcinoma that demonstrates either supraglottic or subglottic spread and/or impaired mobility of the vocal fold. This broad definition has resulted in the inclusion of heterogeneous lesions into one prognostic category. Accordingly, treatment of T2 glottic carcinoma as reported in the literature has yielded heterogeneous results.

With the present study, we aimed to assemble and analyze the existing data that describe the influence of vocal fold mobility on the treatment outcomes of T2 glottic carcinoma. We expected that the current staging system for glottic carcinoma might be inadequate with regard to predicting prognosis in T2 lesions when impaired vocal fold mobility is present. As such, we sought to use meta-analysis techniques to summarize the existing data on this subject.

A search of the English-language literature was conducted in the MEDLINE database via the online Ovid portal for publication dates from January 1, 1950, through September 30, 2007. A predetermined set of inclusion criteria was used. Search terms included laryngeal neoplasm, carcinoma, glottis, and glottic carcinoma. The initial search returned 581 results. These articles were evaluated to identify those studies that reported results from treatment of T2 glottic lesions, which yielded 149 results. These studies were then reviewed to identify those that explicitly reported results according to the presence or absence of pretreatment impairment of vocal fold mobility. Additional references were obtained by manual review of citations in published articles. Outcome measures of interest included local control of disease, local failure, ultimate local control of disease after salvage therapy, absolute survival, and disease-specific survival. Studies were included if they reported at least 1 of the outcome measures with regard to vocal fold mobility. In total, 35 articles were identified that met the inclusion criteria.

Fourteen studies were excluded from analysis. Three studies^27-29 that used primary radiotherapy reported outcomes with stratification by vocal fold mobility but included both glottic lesions and other subsites in a single group. One study³⁰ that used primary radiotherapy reported combined outcomes for T2b and T3 lesions. One study³¹ was excluded because only percentages alone without absolute numbers were reported for analysis. Two studies^32,33 were excluded because of degenerate results on odds ratio (OR) calculation. Several studies^22,23,34-39 reported data on patients included in reports published elsewhere. Ultimately, 21 citations were included for analysis (Table 1).

Each study was analyzed as a case-control study, with patients with impaired vocal cord mobility representing cases and those with normal vocal cord mobility representing controls. The presence or absence of control or survival was the measured outcome. The ORs and 95% confidence intervals (CIs) were calculated from reported data for primary outcomes of each study. Where stated, statements of statistical significance between groups were noted. In reporting our data, tumors that presented without impairment of vocal fold mobility were designated T2a, whereas tumors that demonstrated impairment of vocal fold mobility were designated T2b, without regard to degree of superficial tumor extension.

Meta-analysis was conducted using the Mantel-Haenszel fixed-effects model to pool individual study estimates into an overall summary estimate of treatment effect of radiotherapy on T2 lesions with vocal fold mobility vs those without. Heterogeneity testing using the Cochran Q test with 20 df was not statistically significant (Q = 29.73, P = .07). Bias assessment using the Horbold-Egger formula produced a bias value of 0.6333, which was not statistically significant (P = .47), indicating an unbiased sample (Figure 1). Statistical calculations were performed on the Windows platform using a commercially available software program (StatsDirect 2.6.9; StatsDirect Ltd, Cheshire, England).

Twenty-one articles were included for analysis (Table 1). All studies were retrospective in design. Nineteen studies reported on treatment with radiotherapy alone, and 2 reported a dual-arm study of radiotherapy or surgery alone. The study periods ranged from 10 to 30 years, with a mean of 18.6 years. Sample sizes ranged from 23 to 256, with a mean size of 119. Minimum follow-up periods, where stated, ranged from 24 to 60 months. All studies reported either local control or local failure as primary outcome measures. In addition, 7 studies^{2,3,5,8,9,11,20} reported ultimate control after salvage therapy, 5 studies^5-7,9,21 reported absolute survival, and 4 studies^9,11,16,21 reported disease-specific survival.

All articles, with 1 exception,² reported results for 5-year local control of disease (Table 2). The mean (SD) 5-year local control of T2a lesions was 76.2% (10.2%); by comparison, the mean (SD) local control for T2b lesions was 64.4% (10.2%). Calculated ORs for local control in patients with T2a lesions compared with those with T2b lesions ranged from 0.855 to 11.5. Comparison of local control for T2a vs T2b lesions showed a statistically significant result that favored the T2a lesions (OR, 1.83; 95% CI, 1.52-2.20; P < .001) (Figure 2).

A smaller subset of included studies reported data on ultimate survival after salvage surgery, and this was also evaluated by meta-analysis (Table 3). The odds of ultimate control after salvage surgery was significantly better for T2a lesions vs T2b lesions (OR, 1.90; 95% CI, 1.23-2.92; P = .005) (Figure 3).

Absolute survival was reported in 5 studies (Table 4). Comparison of absolute survival in patients with T2a and T2b lesions was not statistically significant (OR, 1.57; 95% CI, 0.93-2.63; P = .09) (Figure 4). Disease-specific survival was reported by 4 studies (Table 5). Analysis of disease-specific survival comparing patients with T2a and T2b lesions showed a statistically significant difference (OR, 2.51; 95% CI, 1.44-4.39; P = .002) (Figure 5).

Nodal disease was rare in the included studies. Two studies^9,17 included a combined total of 3 patients who initially presented with N1 disease; all other patients had stage N0 disease. The remaining 19 studies included only patients without nodal metastasis. There was no report of any patient with distant metastasis at initial presentation.

Differing treatment outcomes for T2 glottic carcinoma with regard to vocal fold mobility were first noted by Mårtensson et al¹ in 1967. In 1974, Wang² reported on a series of 111 cases of laryngeal carcinoma treated with primary megavoltage irradiation in which T2 glottic lesions demonstrated significantly worse local control rates when preoperative vocal fold function was impaired, without fixation, than when vocal fold function was normal. Harwood and DeBoer²² described similar findings for a series of 164 patients with T2 glottic carcinoma and proposed the subdivision into T2a and T2b according to normal or impaired vocal cord mobility, respectively. Subsequent authors^20,36,40 also noted the tendency for local control of so-called T2a lesions to approach that achieved with T1 lesions, whereas local control for T2b lesions approached the results for T3 lesions, which by definition includes vocal fold fixation.

The present study uses meta-analysis techniques to combine previously published data into a summary estimate of effect. These results indicate with 95% confidence that the true population odds of local control by radiotherapy is between 1.52 and 2.20 times greater in patients with T2a lesions compared with T2b lesions. This is generally supported by the results of the individual included studies. Although a few studies^5,18 report no improvement in local control for T2a lesions, most published reports^{1-4,6-17,19-21} indicate a difference.

Our review of the literature reveals that no prospective clinical studies have been reported in the English-language literature to date that assess treatment outcomes of T2 glottic carcinoma with regard to vocal fold mobility. Many authors conducting retrospective studies^{2,11,22-24,26,27} have noted significant differences between the 2 subgroups with respect to clinical course, response to radiation therapy, and local control rates. Currently, most centers report treatment of T2 glottic carcinoma with primary radiotherapy alone, with surgical salvage reserved for persistent or recurrent disease.

The success of salvage surgery has been another area of interest in T2 glottic carcinoma. Several authors noted that T2a lesions that recurred had high success rates of surgical salvage, whereas recurrent T2b lesions tended to fare more poorly despite salvage surgery. Wiggenraad et al²⁹ reported a rate of successful surgical salvage of 88% for T2a lesions compared with 18% for T2b lesions, although that series included a number of supraglottic and subglottic lesions as well. Kaplan et al⁷ reported similar results, with successful salvage surgery in 70% of T2a tumors compared with 11% of T2b tumors. Van den Bogaert et al⁵ reported surgical salvage rates of 81% for T2a lesions and 68% for T2b lesions, whereas Turesson et al¹¹ reported surgical salvage as 83% for T2a and 67% for T2b lesions. These findings are supported by the present study, which indicates an overall statistically significant odds of successful surgical salvage for T2a lesions, with ultimate survival ranges of 1.93 to 2.92. These data suggest that local control of T2b glottic carcinoma may be best obtained by initial treatment with definitive surgery. Potential options for primary surgical management include laryngofissure with cordectomy, endoscopic surgery, and partial vertical hemilaryngectomy with its various modifications and reconstructive options.^41-46

Surgical management provides an alternative to radiation therapy for T2 glottic carcinoma, although few studies have reported outcomes for T2a vs T2b glottic tumors. Kaplan et al⁷ retrospectively compared the results of primary irradiation vs total laryngectomy in 28 patients with T2b glottic carcinoma. Five-year local control rates were reported as 85% for surgery vs 50% for radiation therapy. Turesson et al¹¹ compared results for 33 patients receiving primary partial laryngectomy, with or without preoperative radiotherapy, with 5-year survival of 95% for those with T2a lesions and 100% for those with T2b lesions. Laccourreye et al⁴⁴ reported significant local failure rates for 416 patients undergoing partial laryngectomy for T2 transglottic lesions as 25.6% if vocal cord mobility was normal and 32.2% if mobility was impaired. Because surgical management is varied even within these few series, the ability to draw conclusions is limited.

Absolute and disease-specific survival have been reported by some authors in the present series. A combined analysis of absolute survival in the 5 included studies fails to demonstrate a statistically significant difference. By contrast, combined analysis of disease-specific survival in the 4 included studies suggests a statistically significant difference between T2a and T2b lesions. However, because of the small number of included studies in this analysis, definite conclusions cannot be readily drawn.

Nodal disease with primary T2 glottic carcinoma was generally rare in the reported series. This finding may be attributed to the known scarcity of channels for regional lymphatic spread in tumors confined to the glottis. Similarly, distant metastasis on initial presentation is rare and did not occur in the reported series. This tendency of glottic carcinoma to remain localized until an advanced stage emphasizes the paramount importance of successful control of the primary lesion at the time of initial treatment.

Among the limitations of the present study are those inherent in all meta-analyses, including the potential for heterogeneity among studies and publication bias. However, additional testing of the 21 included studies revealed good combinability, without significant interstudy bias or heterogeneity. Additional limitations arise from the inclusion of retrospective studies, which predisposes the analysis to historical bias and reporting bias. Furthermore, within each study, it was not possible to determine the rationale for receiving radiotherapy vs surgical management, which introduces selection bias. Accurate staging depends heavily on physical examination, including indirect and fiberoptic laryngoscopy, the interpretation of which may be subject to observer bias. In particular, diagnosis of the presence or absence and degree of vocal cord mobility relies on the impression of each examiner, which may vary significantly. Standardized diagnostic modalities may be necessary in the future to ensure an accurate determination of vocal cord function for staging purposes.

Although the tendency to treat T2 glottic cancer with primary radiation has continued, the technical aspects of such treatment have been subject to modification. The period in history during which treatment has occurred may reflect differences in total dose, fractionation, field size, or treatment duration, among other factors. In a study by Franchin et al¹⁸ of 410 patients with T1 or T2 glottic carcinoma treated with radiotherapy between 1986 and 2001, treatment after 1996 had a significant positive effect on disease-free survival by univariate and multivariate analysis. A recent study by Frata et al²⁰ included 256 consecutive patients with T2N0 glottic carcinoma treated with primary radiation during a 3-decade period from 1970 to 1999. Univariate analysis by decade revealed no significant difference in local control at 3, 5, or 10 years. Similarly, no significant difference was found with regard to total dose, beam type, treatment volume, field size, treatment duration, or fraction dose.²⁰ Whether these and other treatment variables have affected the outcomes of T2a vs T2b glottic lesions remains to be fully studied.

As technology has advanced, many centers have implemented changes in radiation administration schemes. Early studies^{1,2,4-6,10,13} typically reported results obtained with use of cobalt 60 teletherapy, whereas more recent studies^8,12,33 have reported on photon or electron therapy using a linear accelerator, and some^{9,11,15,17,20,32} have used both sources. Radiotherapy with a linear accelerator is well established as superior to cobalt 60.⁴⁷ Recent studies have shown improved outcomes for T2 glottic carcinoma with shortened treatment duration,³⁰ increased total dose,⁴⁸ and twice-daily fractionation schemes.¹⁹ However, published reports addressing the effect of treatment differences on outcomes with regard to T2a vs T2b lesions are currently lacking. As such, whether these differences can account for the findings of the present study remains to be determined.

The addition of chemotherapeutic agents to treatment regimens for glottic cancer has been studied in randomized trials as induction chemotherapy before radiation^49-51 and as concurrent chemoradiotherapy.^52,53 Recently, there has been interest in the application of chemotherapy as a radiosensitizer to improve local control for early glottic lesions.⁵⁴ Current guidelines published by the American Society of Clinical Oncology recommend concurrent chemoradiotherapy for selected patients with T2 glottic cancer and node-positive disease for whom total laryngectomy is the only surgical option or when the functional outcome after larynx-preservation surgery is expected to be unsatisfactory.⁵⁵ The use of chemotherapy in organ preservation protocols, specifically for T2 lesions, is currently under investigation.^56-58 Although initial reports appear promising, the use of radiotherapy alone for treatment of T2 lesions remains widespread, and the effect on outcomes of T2a vs T2b lesions remains unknown.

In view of the negative impact of vocal fold immobility on local control by radiotherapy and considering favorable results with respect to voice quality, swallowing function, and length of treatment, laryngeal preservation surgery may be considered as a therapeutic option.^11,22-24,43 A prospective study of T2a and T2b glottic carcinoma treated with primary surgery is necessary to define the expected benefits of such an approach. Prospective, head-to-head comparison of surgical management with radiotherapy would provide additional valuable information. Until the results of such studies are known, partial laryngeal surgery should be offered as an alternative to radiation for patients with T2b glottic carcinoma.

The results of the present study support previous findings that indicate that impaired vocal fold mobility has a negative impact on local disease control regardless and independent of the presence or absence of extension beyond the vocal fold.^{2,4,6,7,14,15,18} On the basis of these findings, we agree with the designation of T2b for glottic carcinoma with impaired vocal cord mobility, whereas the designation of T2a is reserved for lesions with supraglottic or subglottic extension but with fully mobile vocal cords. Moreover, continued use of the term early-stage disease to describe T2b lesions may risk erroneous inclusion in a better prognostic category, with subsequent application of suboptimal therapy. Future revisions of the American Joint Committee on Cancer staging guidelines may consider adding a formal subdivision of glottic carcinoma into T2a and T2b. Formal incorporation into the Cancer Staging Manual would reinforce the prognostic significance apparent in the literature to date.

Correspondence: Gady Har-El, MD, 186 E 76th St, New York, NY 10021.

Submitted for Publication: September 1, 2008; final revision received October 23, 2008; accepted October 27, 2008.

Author Contributions: Drs McCoul and Har-El had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: McCoul and Har-El. Acquisition of data: McCoul. Analysis and interpretation of data: McCoul and Har-El. Drafting of the manuscript: McCoul. Critical revision of the manuscript for important intellectual content: McCoul and Har-El. Statistical analysis: McCoul. Study supervision: Har-El.

Financial Disclosure: None reported.

Previous Presentation: This study was presented at the Seventh International Conference on Head and Neck Cancer; July 20, 2008; San Francisco, California.