O’Shea R, Byrne H, Tuckett J, O’Leary G, Sheahan P. Impact of Current Smoking and Alcohol Consumption on Gastrostomy Duration in Patients With Head and Neck Cancer Undergoing Definitive Chemoradiotherapy. JAMA Otolaryngol Head Neck Surg. 2015;141(5):463-469. doi:10.1001/jamaoto.2015.0279
Prophylactic gastrostomy tube (GT) insertion prior to chemoradiotherapy is a valuable nutritional adjunct in patients with head and neck cancer undergoing nonsurgical treatment. However, concerns have been expressed regarding high rates of GT dependence. There is little information in the literature regarding the impact of modifiable risk factors such as smoking and alcohol consumption on duration of GT use and dependence rates.
To study the duration of GT use in patients with head and neck cancer at our institution and to investigate risk factors for prolonged GT duration and dependence, including the impact of smoking and alcohol consumption.
Design, Setting, and Participants
This retrospective study at an academic teaching hospital included 104 patients with squamous cell carcinoma of the head and neck undergoing definitive treatment with either chemoradiotherapy (n = 84) or radiotherapy alone (n = 20).
Prophylactic GT insertion prior to commencement of treatment.
Main Outcomes and Measures
Duration of GT use after completion of chemoradiotherapy or radiotherapy and the impact of risk factors on same. Duration of GT use was analyzed using Kaplan-Meier survival curves, with censoring of patients who developed cancer recurrence or died.
The median duration of GT use was 9 months. The actuarial GT persistence rate at 1 year was 35%. Current smoking (hazard ratio [HR], 0.47; 95% CI, 0.27-0.81; P = .01) and current heavy alcohol consumption (HR, 0.55; 95% CI, 0.32-0.97; P = .04) were significant predictors of GT persistence. On multivariate analysis, only current smoking remained significant (HR, 0.53; 95% CI, 0.30-0.94; P = .03). Age 65 years or older, advanced T classification, larynx and/or hypopharynx primary site, and posttreatment neck dissection were not significant risk factors.
Conclusions and Relevance
Current smoking and current heavy alcohol consumption are predictive of prolonged GT requirement in patients with head and neck cancer undergoing prophylactic GT insertion before definitive radiotherapy or chemoradiotherapy. Further study is required to investigate whether smoking or drinking cessation can reduce GT dependence rates in this population.
Definitive chemoradiotherapy has become well established as a means of treatment of advanced cancer of the head and neck, allowing organ preservation with satisfactory oncological outcomes.1 Among the major concerns regarding chemoradiotherapy are acute toxic effects and long-term adverse functional sequelae. In the acute setting, mucositis, xerostomia, and odynophagia may compromise nutritional intake, leading to weight loss, malnutrition, and in severe cases, dehydration and the need for treatment breaks.1 In the long term, late sequelae of chemoradiotherapy including fibrosis of neck and pharynx may lead to chronic dysphagia and aspiration.2
To obviate issues relating to inadequate oral intake in the acute setting, many institutions recommend prophylactic gastrostomy tube (GT) insertion prior to commencement of treatment, which has been shown to reduce weight loss3- 6 and hospital admissions during treatment.3,5,7,8 Regarding long-term outcomes, the best protection against chronic dysphagia would appear to be continuation of swallowing and oral dietary intake during and after treatment,9,10 with swallowing exercises also appearing to confer protective effect.9,10
Smoking and alcohol consumption are well-established risk factors for causation of head and neck squamous cell carcinoma (SCC). Smoking and alcohol use are also known to compromise oncological efficacy of nonsurgical treatment.11- 13 Less is known regarding the impact of continued smoking and alcohol consumption during treatment on functional outcomes.
At our institution, prophylactic GT placement is recommended for nearly all patients undergoing concurrent chemoradiotherapy. The purpose of the present study was to investigate duration of GT use and rate of GT dependence among patients at our institution and to examine risk factors for increased GT use duration. In particular, we wished to study the impact of smoking and alcohol consumption, as potentially modifiable lifestyle risk factors, among other factors, on this outcome.
We undertook a retrospective review of cases of patients who underwent prophylactic GT insertion prior to definitive chemoradiotherapy or radiotherapy for head and neck cancer. At our institution, the general policy was to recommend prophylactic GT insertion prior to commencement of treatment for all patients with head and neck cancer undergoing concurrent chemoradiotherapy as well as for selected patients undergoing radiotherapy alone. Cases for the present study were identified from a database of all patients with GTs maintained by the Nutrition and Dietetics Department. Inclusion criteria were (1) newly diagnosed SCC of the oral cavity, oropharynx, nasopharynx, hypopharynx, or larynx; (2) definitive primary radiotherapy or primary concurrent chemoradiotherapy, with curative intent; and (3) prophylactic GT insertion prior to commencement of treatment between the years 2009 and 2012, inclusive. Exclusion criteria were therapeutic GT placement after initiation of treatment; SCC from nonmucosal head and neck sites or disease other than SCC; history of head and neck cancer or radiotherapy to the head and neck region; primary surgical treatment; or treatment with palliative intent.
Case notes of included patients were reviewed. The data collected included demographic details; details of tobacco and alcohol consumption; site, stage, and histologic characteristics of the primary cancer; details of treatment administered; duration of GT use; weight gain or loss during period of follow-up; oncological outcomes; and death (Table 1).
For the purpose of the study, patients’ alcohol consumption was divided into the following categories: nondrinkers; moderate drinkers; heavy drinkers; and former heavy drinkers. The heavy drinker category was used for patients who drank alcohol every day, patients who drank in excess of 15 (women) or 20 (men) units per week, or patients with history of alcoholism or alcohol-related illness who were currently still drinking. Moderate drinkers included all other current drinkers. Former heavy drinkers were patients who had a history of heavy alcohol consumption but who had not consumed any alcohol in the preceding 1 year.
Radiotherapy was administered by conventional techniques with standard fractionation (2 Gy/d). Chemotherapy was generally administered for 6 weeks as weekly cisplatin or carboplatin with fluorouracil administration.
Patients treated with chemoradiotherapy were followed up regularly in head and neck clinics, including assessments by speech and language therapists and dieticians. Patients were encouraged to maintain a full oral diet without using the GT during and after treatment. In addition, all patients were instructed on swallowing exercises and told to practice these daily, regardless of whether their swallowing was impaired. Gastrostomy tube feedings were prescribed for patients unable to achieve adequate oral intake or maintain weight.
The decision to remove a GT was made by the head and neck surgeon in agreement with the dietician. Tubes were generally recommended for removal in patients who did not require tube feedings in the preceding couple of weeks and were maintaining adequate oral intake and weight.
Patients were followed up until the date of GT removal; date of death; or date of first locoregional recurrence of head and neck cancer with GT still in situ. Patients were censored once they developed any locoregional or distant recurrence of their head and neck cancer because in those cases, the reasons for GT maintenance might well have been to address recurrent cancer effects on swallowing and nutritional status (or in anticipation of same) rather than to address sequelae of initial chemoradiation treatment.
Statistical analysis was performed using XLSTAT, version 2013.5.05 (Addinsoft). Duration of GT use was analyzed by survival analysis using the Kaplan-Meier method. Among patients not undergoing GT removal, cases were censored at the date of last follow-up, date of death, or date of locoregional recurrence of cancer with GT still in situ. Any patient developing locoregional or distant recurrence of head and neck cancer while the GT was still in situ was censored on the date of diagnosis of this recurrence, generally the date of the biopsy confirming the recurrent cancer or the date of radiology study showing recurrent cancer in patients not undergoing biopsy.
A further distinction was made between patients for purposes of censorship. Patients with cervical nodal disease at the outset of treatment who had an incomplete response in the neck to chemoradiotherapy and who required posttreatment neck dissection were not considered to have locoregional recurrence, and these patients were not censored. However, patients with initial complete response in the neck who developed subsequent regional recurrence were considered to have locoregional recurrence and were censored on the date of diagnosis of recurrent nodal disease.
Univariate analysis of impact of variables under study on duration of GT use was performed using Cox regression analysis. Multivariate analysis was performed using Cox proportional hazards modeling with variables included in the model at univariate P < .10. Statistical analysis of 2 × 2 contingency tables in the analysis of factors affecting weight loss was performed using the Fisher exact test.
Ethics approval was obtained from the Cork Clinical Research Ethics Committee, waiving written informed consent.
The final study population consisted of 104 patients (82 men) with a mean age of 60.2 years (age range, 32-84 years). Primary tumor sites were oral cavity (n = 7), oropharynx (n = 46), nasopharynx (n = 4), hypopharynx (n = 27), and larynx (n = 20). Sixty-one patients had T3 or T4 primary tumors; 74 had clinically involved cervical nodes; and 62 had AJCC/UICC stage IV cancers (American Joint Committee on Cancer/Union for International Cancer Control). Sixty-one patients were current smokers; 31 were former smokers; and 12 were never smokers. Thirteen patients were nondrinkers; 31 were moderate drinkers; 44 were heavy drinkers; and 16 were former drinkers (of whom 12 had a history of heavy drinking).
The median radiation dose administered was 70 Gy. Three patients stopped undergoing treatment before completing the prescribed course of radiotherapy after receiving doses of 20 Gy, 30 Gy, and 50 Gy, respectively. One of these was considered to have persistent disease at the time treatment was stopped and was censored at that time. The other 2 did not have clinical evidence of disease at the time treatment was stopped but developed subsequent recurrence and were censored at 8 and 13 months, respectively.
Chemotherapy was administered concurrent with radiotherapy in 84 patients. Forty-nine patients received carboplatin with fluorouracil; 28 received cisplatin with fluorouracil; and 7 received cetuximab. All chemotherapy was administered on a weekly basis, with 6 weeks of treatment planned. Twenty-six patients (15 receiving carboplatin, 10 receiving cisplatin, and 1 receiving cetuximab) did not complete the entire chemotherapy schedule (6 rounds) owing to toxic effects. The median and mean number of weeks of chemotherapy completed were 6 and 5.3, respectively (range, 1-8 weeks).
Twelve patients underwent neck dissection for incomplete response in the neck at a median of 3.5 months (range, 1.5-8.0 months) after completion of treatment. In 4 of these, the GTs had already been removed. Thus, the effect of posttreatment neck dissection was studied in 8 patients with GTs still in situ at the time of neck dissection. Table 1 includes patient demographics, tumor site and stage, and treatment administered.
Local recurrence developed in 33 patients; regional recurrence in 10 patients; and distant recurrence in 14 patients. Median time to first recurrence in patients with recurrence was 7 months (mean, 10.7 months).
The median time from completion of treatment until censoring was 8 months (mean, 10.0 months). For patients who did not undergo GT removal, die, or experience locoregional recurrence of cancer, the median and mean follow-up times were 11 months and 17.3 months, respectively (range, 5-57 months).
The median duration of GT use after completion of treatment was 9 months (mean, 9.8 months). Gastrostomy tubes were removed in 60 patients. Thirty-one patients died with GTs in situ, and 13 had GTs still in situ at their most recent follow-up. Sixty-three patients (61%) still had GTs in situ 6 months after completion of treatment, and 33 (32%) still had GTs in situ 12 months after completion of treatment. The 6- and 12-month actuarial rates of GTs were 60% and 35%, respectively. The 12-month actuarial GT rates according to primary tumor site were 33% (oropharynx), 36% (hypopharynx), 47% (larynx), 50% (oral cavity), and 25% (nasopharynx).
Table 2 summarizes the univariate analysis of predictive factors for “GT survival” (GT use duration). Only current heavy alcohol consumption (hazard ratio [HR], 0.55; 95% CI, 0.32-0.97; P = .04) and current smoking (HR, 0.47; 95% CI, 0.27-0.81; P = .01) were significant. Age 65 years or older, primary tumor site, T3 or T4 classification, N1 to N3 classification, addition of chemotherapy, and performance of posttreatment neck dissection were not significant risk factors (Table 2).
A multivariate analysis with the input variables current smoking and current heavy drinking was performed. On multivariate analysis, only current smoking remained significant (HR for GT removal, 0.53; 95% CI, 0.30-0.94; P = .03).
We repeated the analysis of predictive factors for GT survival on the subset of 77 patients who were treated with concurrent radiotherapy and platinum-based chemotherapy (excluding patients treated with radiotherapy alone or concurrently with cetuximab). The results of the univariate analysis are listed in Table 3. Current smoking (HR, 0.42; 95% CI, 0.23-0.77; P = .01) and current heavy alcohol consumption (HR, 0.53; 95% CI, 0.29-0.99; P = .047) were once again significant. Cancer T3 or T4 classification was also significant (HR, 0.53; 95% CI, 0.29-0.96; P = .04). The impact of age 65 years or older did not reach significance (HR, 0.55; 95% CI, 0.30-1.01; P = .06).
Multivariate analysis was performed with the input variables current smoking, current heavy drinking, T3 or T4 classification, and age 65 years or older. Once again, only current smoking was significant (HR, 0.51; 95% CI, 0.27-0.97; P = .04) (Table 3).
Ninety-three patients had documentation of weight before and 6 months after completion of treatment. Sixteen of these had recurrence of their cancer within 6 months and were thus excluded from analysis of risk factors for weight loss during and after treatment. Mean (SD) weight loss among the remaining 77 patients was 3.8 (6.8) kg. Thirty-nine of these patients (51%) experienced greater than 5% weight loss, and 22 (29%) experienced greater than 10% weight loss.
Table 4 summarizes the univariate analysis of risk factors for greater than 5% weight loss. Cancer N1 to N3 classification (P = .04), and posttreatment neck dissection (P = .01) were significant. Weight loss was also highly significantly associated with GT being maintained in situ at 6 months (P = .01).
Percutaneous GTs provide invaluable nutritional support for patients with head and neck cancer undergoing radiotherapy and chemoradiotherapy and have been associated with less weight loss during treatment,3- 6 fewer hospitalizations,3,5,7,8 and fewer treatment interruptions.3 Concerns have been raised by some authors that prophylactic GT insertion increases duration of GT use,4,6,14,15 increases late dysphagia,6,14,15 and leads to increased likelihood of GT dependence.6,15 This is probably attributable to patients with GTs in situ making less effort to swallow, leading to increased fibrosis and disuse atrophy of pharyngeal muscles. However, other authors have reported prophylactic GTs to be associated with lower rates of stricture8 and aspiration8 and not associated with higher GT dependence rates.5,8
At our institution, the general policy is to recommend prophylactic GT insertion for all patients with head and neck cancer undergoing definitive treatment with chemoradiotherapy and for selected patients undergoing radiotherapy alone who are considered to be at high risk of dysphagia. In the present series, we report a high incidence of persistent GT presence at 6 and 12 months after completion of treatment. The main factors that appeared to impact GT use duration were current smoking and current heavy alcohol consumption, with only current smoking significant on multivariate analysis. The major risk factors for greater than 5% weight loss were N1 to N3 nodal classification and performance of posttreatment neck dissection.
In analyzing our data, we studied the duration of GT use and impact of variables on same by survival analysis using the Kaplan-Meier method and Cox regression modeling. This is in contrast to studies that have reported prevalence of GTs at a given time point after completion of treatment.16 We believe survival analysis to be a more accurate way to study duration of GT use than examination of the prevalence of persistent tubes at a given time point because it avoids bias due to early exclusion of patients developing recurrence or becoming lost to follow-up. In addition, our patients were censored at first development of any recurrence. The reason for this is that we wished to investigate factors predictive of prolonged GT duration in patients undergoing successful treatment of head and neck cancer with radiotherapy or chemoradiotherapy. In patients developing recurrence, the reasons for maintaining GTs in situ might relate to dysphagia or nutritional insufficiency caused by the recurrent tumor, or anticipation of same, rather than factors relating to the initial treatment. In contrast, other studies have reported cancer recurrence as the only independent predictor of GT use duration.5
The prevalence of GTs at 1 year in the present study is comparable to a GT prevalence of 29% at 2 years reported by Machtay et al16 in a widely cited article that reported on the analysis of patients from a subset of 3 previously reported Radiation Therapy Oncology Group (RTOG) trials. Among other studies, typically reported GT prevalence at 1 year after completion of chemoradiotherapy ranges from 6% to 26%.6,8,10,17,18 Reasons for this variability are likely multifactorial, including differences in patient populations, radiotherapy techniques and schedules, chemotherapy types and schedules, policies regarding GT placement (prophylactic vs therapeutic) and removal, degree of support by nutritionists and speech and swallowing therapists, completeness of follow-up and proportions of cases excluded, and how data were analyzed.
The variables that we found to be predictive of prolonged duration of GT use on univariate analysis were current smoking and current heavy alcohol consumption, with only current smoking remaining significant on multivariate analysis. Impact of smoking and alcohol consumption on functional outcomes and GT duration in patients with head and neck cancer undergoing nonsurgical treatment has been little reported in the literature. In a very recent article, Bhayani et al10 reported current smoking to significantly impact GT duration in patients with hypopharyngeal cancer treated nonsurgically, while Haderlein et al19 reported alcohol abuse to have adverse impact on posttreatment swallowing function in patients with laryngopharyngeal cancer treated with induction chemotherapy followed by definitive chemoradiotherapy.
The reasons for effect of smoking and alcohol consumption are probably multifactorial. Smoking has been reported to increase the severity of mucositis,20- 22 while nicotine may suppress appetite, leading to less effort by the patient to resume full oral diet. Nutritional deficits are common in heavy drinkers,23 which may have mandated a longer duration of GT use. Another possibility is that current smoking and heavy alcohol consumption may be surrogates for poor performance status, poor patient motivation, or both, and that these were the primary reasons for failure to resume full oral diet after completion of treatment.
The finding of significant impacts for smoking and alcohol consumption on GT duration raises the question whether smoking and drinking cessation prior to treatment may lessen the risk of long-term GT use. Support for this notion may come from our findings that only current smoking and heavy alcohol consumption were risk factors for GT persistence. However, given the potential for bias due to other variables not controlled for in the present study, including performance status, socioeconomic class, and other factors, this theory would need to be validated in a prospective study. Of note, Zevallos et al20 reported a lower incidence of mucositis and pharyngeal strictures in smokers with head and neck cancer who ceased smoking prior to radiotherapy than in continuing smokers, but the differences were not significant owing to small numbers of patients. Nevertheless, given the established benefits of smoking and drinking cessation in patients undergoing treatment for head and neck cancer, it would seem quite reasonable to advise patients to endeavor to abstain from these behaviors during their treatment.
In the study by Machtay et al,16 significant factors for severe late toxic effects after chemoradiotherapy were T3 or T4 primary tumor classification, larynx and/or hypopharynx primary site, older patient age, and neck dissection after chemoradiotherapy. Several of these factors have also been reported by other authors to be predictive of the need for GT use.15 We did not find any of these factors to be significant in the overall group. However, when the analysis was restricted to patients undergoing chemoradiotherapy with concurrent platinum-based chemotherapy, T3 or T4 primary tumor classification was significant for longer GT use, and age 65 years or older also increased the duration of GT use, although the difference was not statistically significant. In contrast, we did not find any significant impact for primary tumor site or neck dissection after radiotherapy, although neck dissection after chemoradiotherapy was associated with a significantly higher risk of weight loss greater than 5% at 6 months.
Our study differs from that of Machtay et al16 and from other earlier studies in several important ways that might explain our different results. First, GTs were by policy inserted prophylactically in all patients undergoing primary chemoradiotherapy in our series. In contrast, studies in which most of the patients underwent therapeutic GT insertion may be subject to bias caused by inclusion of a greater proportion of cases with incipient swallowing compromise due to the primary tumor itself.
Second, one of the drawbacks of the study by Machtay et al16 and others was the large numbers of patients excluded, mostly owing to having recurrent cancers or being lost to follow-up. In the present study, we circumvented this problem by analyzing data according to survival analysis, obviating the need to exclude these cases.
Finally, it is possible that there were differences between study populations due to local treatment policies. For example, at our institution, primary surgery is generally considered for all patients with oral cavity cancer and for many patients with “favorable” oropharyngeal cancers, where a good outcome from transoral laser microsurgery is anticipated. Thus, it is possible that patients with oral cavity or oropharyngeal SCC in the present series had a greater preponderance of less favorable tumors that were more likely to be infiltrating deep muscle, and so these patients were more likely to develop treatment-related dysphagia. Primary surgery was also considered for many cases of advanced bulky laryngeal and hypopharyngeal cancers, which may have similarly introduced a bias in the cohort of patients with laryngohypopharyngeal cancer included in our study. These differences in study populations may be a major reason for the lack of impact of primary tumor site on GT duration in our series.
We did not find the addition of chemotherapy to be a significant risk factor for prolonged duration of GT use over radiotherapy alone. Since it was our policy to perform prophylactic GT insertion only in selected patients undergoing radiotherapy alone, the reasons for those patients being selected (anticipated dysphagia owing to patient or tumor characteristics) likely biased our analysis.
Weaknesses of the present study include the retrospective nature, with reliance on documentation within medical records for information regarding smoking status and alcohol consumption history, and the absence of prospectively collected performance status scores. A further issue is the lack of a control group of patients undergoing radiotherapy or chemoradiotherapy but not undergoing prophylactic GT insertion. We also did not control for size of treatment fields, which may also affect severity of mucositis.24 Related to this, patients in the present study were treated by conventional radiotherapy techniques, not by intensity-modulated radiotherapy (IMRT), which allows more precise treatment planning and has been shown to reduce xerostomia by parotid gland sparing over conventional techniques.18,25 It is possible that IMRT may also lead to better swallowing outcomes and less likelihood of GT dependence, although this is an area that is still under study.18,26,27 Finally, duration of GT use after completion of chemoradiotherapy or radiotherapy alone is a crude surrogate for severity of posttreatment dysphagia because other factors, including general nutritional status and patient motivation, may also affect the timing of gastrostomy tube removal.
The major strengths of our study are the consistent management of patients with GTs. All GTs were placed, managed, and followed up in our department; all patients were instructed on swallowing exercises; and consistent policies were applied to indications for placement and removal of the tubes. Our analysis of data by survival analysis obviated the need to exclude patients with recurrence, and patients who experienced recurrence were censored at the time the recurrence was discovered. Finally, we examined multiple other variables with putative effect on duration of GT use in addition to smoking and alcohol consumption.
Our results demonstrate a high rate of GT dependence after completion of chemoradiotherapy or radiotherapy in patients with head and neck SCC who had undergone prophylactic GT insertion. Risk factors for prolonged GT duration included current smoking and current heavy alcohol consumption, with only current smoking significant on multivariate analysis. The impact of ever smoking and history of heavy alcohol consumption prolonged GT duration, but the differences did not reach statistical significance. Our results would support advising patients with head and neck SCC undergoing radiotherapy or chemoradiotherapy to avoid smoking and excess alcohol consumption during treatment. However, to determine whether stopping smoking and drinking can shorten duration of GT use will require further data from prospective studies.
Submitted for Publication: November 26, 2014; final revision received January 14, 2015; accepted February 6, 2015.
Corresponding Author: Patrick Sheahan, MB, MD, FRCSI (ORL-HNS), South Infirmary Victoria University Hospital, Cork, Ireland (email@example.com).
Published Online: March 19, 2015. doi:10.1001/jamaoto.2015.0279.
Author Contributions: Dr Sheahan had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors had access to the entire data, and all accept accountability for accuracy of data.
Study concept and design: Tuckett, Sheahan.
Acquisition, analysis, or interpretation of data: O’Shea, Byrne, Tuckett, O’Leary, Sheahan.
Drafting of the manuscript: Tuckett, O’Leary, Sheahan.
Critical revision of the manuscript for important intellectual content: O’Shea, Byrne, Tuckett, Sheahan.
Statistical analysis: Sheahan.
Administrative, technical, or material support: Byrne, Tuckett.
Study supervision: O’Leary, Sheahan.
Final approval of the article: O’Shea, Sheahan.
Conflict of Interest Disclosures: None reported.
Additional Contributions: We thank Ann Hevers, BSc, and Linda Perrott, BSc, Nutrition and Dietetics Department, South Infirmary Victoria University Hospital, for maintenance of our database of gastrostomy patients.