A Half-Smile: Dental Health and Survival After Esophagectomy

As the population grows increasingly accustomed to having highly accurate predictive models at its fingertips for everything from the impact of weather on flight delays to odds for sports betting, there are growing expectations for the medical field to sophisticate and simplify our predictive algorithms at a comparable pace. However, we know this is no easy feat in the case of predicting prognosis, adverse drug effects, or surgical outcomes. In the case of esophagectomy for curative oncologic treatment, the procedure is highly morbid, with an overall 5 year survival rate as low as 30%.¹ Ability to detect robust risk factors that allow improved short- and long-term outcome prediction could allow us to provide our patients with more confident guidance and also may improve overall outcomes by informing patient selection and identifying opportunities to intervene.

Whereas eyes are considered a window to the soul, some consider the mouth to be a window into a person’s health. As such, oral health has been raised several times in the esophagectomy outcomes literature as both a direct and an indirect variable. In the only prospective study on the topic, Akutsu et al.² showed that perioperative teeth-brushing effectively reduced postoperative pneumonia and the need for tracheostomy after esophagectomy in Japan. Less intuitive than the problems oral pathogens can create in and around a complex surgical field, others also have described impacts on long-term cancer outcomes. The oral bacterium, Porphyromonas gingivalis, has been shown in pre-clinical and clinical models to portend a worse prognosis in squamous cell carcinoma of the esophagus despite optimal therapy by promoting tumor progression and chemoresistance.^3,4

Others have looked to oral health as a marker for overall health and likelihood of medical interactions, akin to socioeconomic status. In two retrospective studies of patients undergoing esophagectomy from Japan, Miura et al.⁵ found that a loss of more than seven teeth shown on a preoperative dental exam was significantly associated with overall survival and progression-free survival, whereas Tamagawa et al.⁶ demonstrated that the Oral Health Assessment Tool score, a multi-criteria oral health survey, was an independent predictor of 5-year overall survival and postoperative pneumonia. The small sample sizes of these single-institution studies precluded their generalizability but opened the stage for further investigation.

In this issue of Annals of Surgical Oncology, Buchli et al.⁷ report on the first and largest multi-institutional population-based cohort study to evaluate tooth loss as a prognostic indicator after esophagectomy for esophageal cancer. Their findings contrasted with prior results and showed no significant association between amount of tooth loss and overall survival. Completion of this study in a Western nation with detailed longitudinal public health data not only provided strength for the study in its large sample size, but also captured an impressive 98% uptake of all patients who underwent esophagectomy in Sweden since 1987, thus ensuring proper representation of the population in question. The depth and detail of the statistical models reflect the careful consideration that went into the design and effectively minimized bias throughout.

Despite the widely cast net reaching nearly the entire target cohort, the authors included only patients with dental exams in the last 5 years, limiting their breadth by nearly half. This may have added potential for selection bias and data inaccuracies and may have missed a major opportunity to comparatively analyze the group against the other half of patients undergoing esophagectomy who did not have available dental data in the nationalized dental registry. Because this was in effect a negative study, comparing patients undergoing esophagectomy and stratifying them by whether they had or did not have dental data also should have demonstrated no difference in outcomes because tooth loss was not associated with outcomes.

Although effort was made to adjust for socioeconomic status using education as a surrogate, it is conceivable that seeing a dentist also is related to social standing as well the likelihood of seeing a doctor for other health issues. And by that logic, aside from the assumption that those patients may have worse dental health, they too could have experienced worse outcomes for other reasons. Even if the group proved those assumptions unfair and false by having comparably intact teeth, comparing those with and without dental data may have provided meaningful analysis to determine whether outcomes were affected, and support for that null hypothesis would have made for an even stronger paper. Moreover, to be complete, we must mention that a dental exam can change quite a bit with a lag time of 5 years, so this may not have been the most accurate measure. Considering that standard anesthesia documentation should typically include preoperative dental findings, which could confirm the data from the dental registry, mining that data during review of medical records may have provided for a more complete analysis.

Buchli et al.⁷ designed and executed a meticulous, comprehensive review that used data linkage of the exquisite detail encompassed by incorporation of various population health data registries in Sweden. Although their analysis was robust and provided valuable findings, the significant amount of data that should have been available from both the dental registry and anesthesia documentation at the time of esophagectomy may have resulted in including more patients. Thus, by excluding such a sizable portion of patients who had no dental reports in the registry in the last 5 years, they based their analysis on potentially outdated information and may have introduced several sources of inadvertent selection bias. If the purpose of the paper is to state that surgical outcomes for esophagectomy are not associated with dental health and that we as surgeons should not be biased based on a patient’s smile, we would like to see this otherwise attractive study repeated with one change: simply counting everyone’s teeth. Then when the patient smiles, we can smile right back.