糖心vlog

Key Points

Question听 Can olfactory dysfunction (OD) be used as an accessible marker of frailty in patients with head and neck cancer?

Findings听 In this prospective cohort study of 51 adult patients with head and neck cancer, psychophysical OD but not self-reported OD was common and associated with increasing levels of frailty and postoperative length of stay.

Meaning听 Olfactory testing may represent a novel and biologically robust screening method to identify individuals with head and neck cancer who are frail in a way that is rapid, objective, easy to interpret, accessible to large populations, and comparable across institutions.

Importance听 Olfactory dysfunction (OD) is increasingly recognized as a robust marker of frailty and mortality. Despite broad recognition of frailty as a critical component of head and neck cancer (HNC) care, there is no standardized frailty assessment.

Objective听 To assess the prevalence of OD and its association with frailty and postoperative outcomes in HNC.

Design, Setting, and Participants听 In this prospective cohort study with enrollment between February 17, 2021, to September 29, 2021, at a tertiary academic medical center, 85 eligible adult patients with primary, treatment-naive HNC of mucosal or cutaneous origin were included. Patients with a history of COVID-19, neurocognitive, or primary smell/taste disorders were excluded.

Exposures听 Prospective olfactory assessments (self-reported, visual analog scale [VAS] and psychophysical, University of Pennsylvania Smell Identification Test [UPSIT]) with concurrent frailty assessment (Risk Analysis Index [RAI]) were used. Olfactory-specific quality of life (QOL) was examined with brief Questionnaire of Olfactory Disorders鈥揘egative Statements (QOD-NS).

Main Outcome(s) and Measure(s)听 The primary outcome was the prevalence of OD as assessed by VAS (0-10, no to normal smell) and UPSIT (0-40, higher scores reflect better olfaction) and its association with frailty (RAI, 0-81, higher scores indicate greater frailty). For surgical patients, secondary outcomes were associations between OD and postoperative length of stay (LOS), 30-day postoperative outcomes, and QOD-NS (0-21, higher scores indicate worse QOL).

Results听 Among 51 patients with HNC (mean [SD] age, 63 [10] years; 39 [77%] male participants; 41 [80%] White participants), 24 (47%) were frail, and 4 (8%) were very frail. Despite median (IQR) self-reported olfaction by VAS of 9 (8-10), 30 (59%) patients demonstrated measured OD with psychophysical testing. No meaningful association was found between self-reported and psychophysical testing (Hodges-Lehmann, <0.001; 95% CI, 鈭�2 to 1); a total of 46 (90%) patients did not report decreased olfaction-specific QOL. Median UPSIT scores were lower in frail patients (Hodges-Lehmann, 6; 95% CI, 2-12). Multivariate modeling demonstrated severe microsmia/anosmia was associated with 1.75 (95% CI, 1.09-2.80) times odds of being frail/very frail and approximately 3 days increased LOS (尾, 2.96; 95% CI, 0.29-5.62).

Conclusions and Relevance听 Although patients with HNC are unaware of olfactory changes, OD is common and may serve as a bellwether of frailty. In this prospective cohort study, a dose-dependent association was demonstrated between increasing degrees of OD and frailty, and the potential utility of olfaction was highlighted as a touchstone in the assessment of HNC frailty.

The dramatic growth of the world鈥檚 population of older adults represents one of the most substantial public health challenges of this era¹ and is accentuated in patients with cancer.² Frailty is a critical concept in each of these populations and is a syndrome, distinct from age, comorbidity, or disability alone, which describes decreased physiological reserve and resistance to stressors.³ Being frail increases vulnerability to poor health outcomes including falls, worsening mobility, difficulties with activities of daily living, disability, hospitalizations, and death.^4-6 In surgical cohorts, including those with head and neck cancer (HNC), frailty is a direct, independent predictor of perioperative complications, morbidity, and mortality.^7-9

Despite the importance of frailty, consensus on defining frailty remains elusive, and no single tool has emerged as the reference standard for measurement. Although most instruments are variations of 2 general approaches, the cumulative deficit^10,11 and the phenotypic^3,12 models, there is no agreement on how to operationalize frailty in clinical practice or for research.^13,14 Most instruments have yet to depict whether targeted interventions translate to sustainable improvements in patient-level and health-systems outcomes.¹ There is emerging evidence that the implementation of frailty screening initiatives may translate to a survival advantage among patients who are frail and undergoing elective surgery.¹⁵ Nevertheless, despite recommendations to incorporate frailty screening into clinical practice,^16,17 there remains a need for quicker, simpler, and standardizable methods to identify individuals who are frail in ways that can support clinical decision-making and health care resource allocation.

The sensorineural function of smell is an emerging novel marker of frailty. Olfactory dysfunction (OD) has increasingly been recognized as a bellwether of frailty and mortality, suggesting that similar to functional performance measures like grip strength and gait speed, smell impairment may serve as a clinical marker of the underlying pathophysiology associated with a decreased physiological reserve and reduced survival.^18-24 There are several hypotheses to describe the association between OD and mortality, including changes in diet, neurodegeneration, senescence, and decreased response to health hazards, among others.²⁵ Meanwhile, a robust body of literature details the substantial negative effects of OD on well-being, quality of life (QOL), depression,^26-29 nutritional status,^30-32 productivity, and cognitive function.^33-36 While some of these observations may account for the association between OD and frailty, the mechanisms between sensorineural dysfunction and downstream disease manifestations remain to be elucidated.

Notwithstanding growing research describing the association between OD and frailty and the implications of frailty in cancer, there have been no previous studies to our knowledge examining the association between OD, frailty, and oncologic outcomes. In line with recommendations by the National Institutes of Aging and a call for pragmatic frailty markers,³⁷ rather than creating a new frailty instrument, we aim to address the unmet need of identifying clinical markers in pretreatment cancer risk assessment by leveraging OD as a biophysiological metric of frailty. The measurement of olfaction is regularly performed in otolaryngology evaluations and offers a cost-conscious, readily accessible clinical marker that benefits from familiarity and repeatability. Furthermore, smell testing potentially overcomes current shortcomings of various frailty assessments including discrepancies in agreement between measures and resource constraints that limit widespread implementation of frailty-focused care pathways.

In this pilot cohort study, we aimed to prospectively evaluate measures of olfaction in patients with HNC. We sought to detect an association between OD and frailty and hypothesized patients with increasing degrees of OD would demonstrate increased frailty. We further compared the utility of self-reported (ie, subjective) olfaction vs psychophysical (ie, objective) olfaction as markers of frailty. Finally, in subset analysis of patients who underwent definitive surgical treatment, we examined the association between OD and frailty as predictors of 30-day postoperative outcomes.

In this prospective cohort analysis of patients with primary HNC at the University of Pennsylvania from February 17, 2021, to September 29, 2021, eligible participants included English-speaking adults aged 18 years and older with previously untreated mucosal or cutaneous malignant neoplasm of the head and neck. These participants were identified and recruited using a clinic-based approach. Participants with a history of previous HNC, prior or concurrent cancer outside of the head and neck, neurocognitive disorders (eg, dementia, traumatic brain injury, Parkinson disease, or Alzheimer disease), previous COVID-19 infection, and/or intrinsic chemosensory pathology were excluded. Written informed consent was obtained with institutional review board approval (protocol No. 844311).

Collected clinical data included primary tumor site, pathology, TNM cancer stage (American Joint Committee on Cancer [AJCC] Staging Manual, eighth edition), and treatment(s) received. For patients treated with definitive surgery, additional outcomes included length of stay (LOS), readmission, unplanned operating room (OR) return, and postoperative complications. Survey data included sociodemographic information, health history, health insurance type/benefit design, frailty assessment, and subjective OD. As financial insolvency has been associated with poorer outcomes and increased mortality,³⁸ financial distress was also evaluated using the validated Financial Distress Questionnaire (FDQ), a 2-item questionnaire (鈥淒oes household income meet basic needs?鈥� [yes/no] and 鈥淗ow difficult is it to pay for basic needs?鈥� [not at all difficult/somewhat difficult/extremely difficult]).³⁹ Severity of financial distress was scored using an ordinal classification scheme (grade 1 [mild], grade 2 [moderate], and grade 3 [severe] distress).³⁹

Frailty was operationalized using the Risk Analysis Index (RAI), a 14-item instrument used for prospective surgical frailty assessment.⁴⁰ The RAI takes patient age, gender, cognition, disability, cancer diagnosis, and comorbidities into account; surveys are scored on a weighted scale, in which higher scores on the index indicate greater frailty. Scores (range, 0-81) were classified into 4 groups: robust (RAI鈥夆墹鈥�29); normal (30鈥夆墹鈥塕AI鈥夆墹鈥�36); frail (37鈥夆墹鈥塕AI鈥夆墹鈥�44); and very frail (RAI鈥夆墺鈥�45).^41,42

Self-reported OD was evaluated with a visual analog scale (VAS), a continuous 10-point slide scale ranging from 0 (no sense of smell) to 10 (normal sense of smell).⁴³ Olfactory-specific QOL was assessed using the brief Questionnaire of Olfactory Disorders鈥揘egative Statements (QOD-NS), which consists of 7 statements reported on a scale of 0 to 3, with higher scores (summed 0 to 21) reflecting worse QOL.⁴⁴ Psychophysical olfaction was measured via the University of Pennsylvania Smell Identification Test (UPSIT, Sensonics Inc), a forced-choice test of 40 microencapsulated (scratch-and-sniff) odors.⁴⁵ In comparison to complex olfactometric tools or odorant bottles or sticks, the UPSIT may be self-administered, decreasing resources required for smell function testing. Each item has 1 correct answer, yielding a maximum score of 40, wherein higher scores reflect better olfaction and a score of 10 (25% correct) reflects random guessing.^45,46 Data were evaluated in relative terms via age-adjusted and sex-adjusted percentiles⁴⁶ and categorized in absolute terms based on previously established and validated UPSIT thresholds⁴⁷: normosmia (鈮�35 in female patients; 鈮�33 in male patients); mild microsmia (31-34 in female patients, 30-33 in male patients); moderate microsmia (26-30 in female patients, 26-29 in male patients); severe microsmia (19-25); and anosmia (6-18).

An a priori power analysis was conducted to determine the minimum sample size required to detect an association between psychophysical OD and frailty status in patients with HNC. Results indicated the required sample size to achieve 80% power at a significance criterion of 伪鈥�=鈥�.05 was n鈥�=鈥�22. Differences in means, Hodges-Lehmann estimates, and 蔚² measures with 95% CIs were used for bivariate comparisons of continuous variables, and differences in proportions and Kendall 蟿 measures were used for categorical variables. Multivariable regression models, adjusted for TNM stage and subsite, were used to characterize the association of olfaction, frailty, LOS, and postoperative outcomes. We elected not to include age, sex, gender, and select comorbidities, as these covariates are included as components of the RAI and UPSIT percentile measures. Analyses were completed using R statistical software, version 4.1.2 (R Project for Statistical Computing).

Among 85 patients screened for eligibility, 51 (60%) met inclusion criteria and were enrolled (eFigure in Supplement 1). Participant characteristics are presented in Table 1; 39 (77%) were male, and 41 (80%) were White, with a mean (SD) age of 63 (10) years. The most common primary tumor subsites included the oropharynx (n鈥�=鈥�26; 51%), oral cavity (n鈥�=鈥�9; 18%), and cutaneous (n鈥�=鈥�5; 10%). Squamous cell carcinoma was the most common tumor pathology (n鈥�=鈥�39; 77%). Most patients presented with localized, stage I to II disease (n鈥�=鈥�31; 61%). No patients were deemed robust, while more than half the cohort was found to be either frail (n鈥�=鈥�24; 47%) or very frail (n鈥�=鈥�4; 8%) by RAI. Of the 51 examined patients, 42 (82%) underwent surgery. No meaningful differences were found in age, gender, UPSIT, or RAI between surgical and nonsurgical groups (Table 1). For 9 patients treated nonsurgically, 4 (44%) were lost to follow-up, and the remaining 5 (56%) were treated with primary definitive therapy (n鈥�=鈥�1 radiation therapy alone; n鈥�=鈥�4 cisplatin-based chemoradiation therapy).

A total of 30 (59%) patients had some degree of measurable OD on psychophysical testing with UPSIT, with more than one-fourth demonstrating severe OD (severe microsomia/anosmia, n鈥�=鈥�12; 24%). However, the median (IQR) self-reported olfaction by VAS was 9 (8-10), and most patients (n鈥�=鈥�46; 90%) did not report baseline impairments in olfactory-specific QOL by the QOD-NS. Table 2 demonstrates notable discordance and no meaningful association between psychophysical olfaction and self-reported olfaction, including VAS responses (Hodges-Lehmann, <0.001; 95% CI, 鈭�2 to 1) and QOL by QOD-NS (蟿, 0.13; 95% CI, 鈭�0.05 to 0.3). Similarly, no meaningful differences were observed by frailty status for olfactory-specific QOL by QOD-NS and VAS responses (Hodges-Lehmann, <0.001; 95% CI, 鈭�1 to 1) (Table 3). However, frailty status was associated with a meaningful decrease in psychophysical OD as measured by median (IQR) UPSIT scores (34 [30-36] vs 27 [22-34]; Hodges-Lehmann, 6; 95% CI, 2-12) (Table 3).

A total of 42 patients (90%) underwent definitive surgical treatment with a mean (SD) LOS of 5 (5) days. Table 4 shows that 9 (21%) were readmitted within 30 days, and 7 (16%) required unplanned OR return. Overall, the postoperative complication rate was 14% (n鈥�=鈥�6: 2 [5%] had hemorrhages; 2 [5%] had wound disruptions; 1 [2%] had a surgical site infection; and 1 [2%] had pneumonia). There was a moderate association between postoperative LOS and olfaction impairment. Median (IQR) days of LOS in patients with normosmia (2 [1-3] days) was shorter as compared with those with mild/moderate microsmia (3 [1-7] days) and severe microsmia/anosmia (6 [4-8] days) (Table 4; 蔚², 0.07; 95% CI, 0.02-0.16). There were no detectable differences in other postoperative outcomes when patients were characterized by olfactory status, including readmission rates, return to the OR, or postoperative complications. Like worse psychophysical olfaction, patients who were frail had increased median (IQR) days of LOS compared with patients who were not frail (3 [1-4] vs 5 [3-8]; Hodges-Lehmann, 2; 95% CI, 0.006-5) without meaningful differences in other untoward postoperative outcomes (eTable in Supplement 1).

Multivariable regression analysis adjusted for TNM stage and subsite, was used to further assess the association between olfaction and frailty with LOS. The odds of being frail/very frail was 1.75 times greater in patients with severe microsmia/anosmia, relative to those with normosmia (RAI鈥夆墺鈥�37: OR, 1.75; 95% CI, 1.09-2.80) (Table 5; see olfaction and frailty). Meanwhile, OD, but not frailty, was a significant predictor of LOS (Table 5; see olfaction and LOS: severe microsmia/anosmia, 尾, 2.96; 95% CI, 0.29-5.62; see frailty and LOS: frail/very frail, 尾, 1.20; 95% CI, 鈭�0.94 to 3.33).

The estimated global prevalence of frailty is 12% to 24% of community-dwelling adults aged 50 years and older,⁴⁸ with higher rates among individuals from lower socioeconomic groups and racial and ethnic minority groups.^49,50 The rapid expansion of the older adult population will parallel increases in the prevalence of frailty, and, as such, systematic frailty identification and management represent a critical public health priority.^1,51 Despite deliberation surrounding conceptual frameworks and tools to measure frailty, there is less debate surrounding the urgent need to detect at-risk individuals before the onset of clinically identifiable frailty, and before taxing medical or surgical interventions that may overwhelm limited physiological reserve. There is evidence to suggest that olfactory impairment may serve as an independent novel marker of frailty. Nevertheless, there have been no previous studies to our knowledge examining the association and implications of OD on frailty in cancer care. In this prospective cohort investigation, we demonstrate that although patients with HNC may be unaware of their OD, diminished psychophysical olfaction represents a unique, dose-dependent, functional performance measure of frailty in this population. Through this pilot study, we also demonstrate that OD may serve as a bellwether of postoperative outcomes. With a broad assessment of olfaction-related measurements, this current report provides the foundation for future study of olfaction as a measure of frailty in patients with cancer.

With escalating scrutiny regarding the delivery of high-value, patient-centered care, assessing the predictive value of frailty concerning morbidity and mortality in oncologic populations is needed.⁵² Traditional oncologic measures, such as the Karnofsky Performance Status and ECOG (Eastern Cooperative Oncology Group) Performance Status scales,^53,54 have been used to assess functional status, but they have not been validated in older adults, are subjective, and do not sufficiently capture information regarding comorbidities, physiological impairment, or discriminate level of risk.^55,56 Though several commonly used frailty tools, such as the Fried frailty phenotype assessment¹² and the Rockwood and Mitnitstki frailty index,⁵⁷ have been examined, the RAI used in this investigation represents the most thoroughly validated prospective measure of surgical frailty and instrument proven feasible for point-of-care, real-time testing to guide preoperative decision making, taking a median 30 seconds to assess without disrupting clinic workflow.^5,40,41 Patients who were frail and very frail as measured by the RAI demonstrate high rates of postoperative mortality across all levels of operative stress,⁵⁸ highlighting that there are no low-stress or low-risk procedures among patients who are frail. Despite the utility and advantages of RAI assessment in the clinical setting, there remains no universally accepted standard for frailty evaluations. The resulting discrepancies in agreement between frailty measures and resource constraints across institutions are some of the challenges that limit the feasibility of widespread implementation of frailty-focused care pathways. In this investigation, we demonstrate that OD is reliably associated with frailty in a population with HNC. The results suggest that smell testing may serve as a biologically robust screening method to identify frail individuals in a way that is objective, easy to interpret, accessible to large populations, and comparable across institutions.

Although most adults will experience a visual, hearing, or olfactory impairment with increasing age, OD is uniquely associated with increased mortality risk.⁵⁹ Olfactory function is one of the strongest predictors of 5-year mortality, and older, anosmic patients demonstrate a 3-fold increased odds of death in comparison to a normosmic control group. Importantly, more than 50% of healthy adults without self-perceived olfactory impairment manifest some degree of occult OD on psychophysical olfactory testing.^21,60 Similarly in this investigation, most patients did not report decreases in smell perception or olfactory-specific QOL, despite 59% of study participants demonstrating measurable deficits on olfactory testing, thus highlighting the importance of psychophysical assessments. Though age is a considerable factor in olfactory function, impaired olfaction in the aged population appears to serve as a unique indicator of physiologic resilience or cumulative pathologic burden. Despite the ostensible risk of OD with age and its potential effect on survival, routine testing of this clinically-accessible sense is uncommon.^60,61

Disruptions in the ability to smell may represent a set of risks uniquely poised to worsen the outcomes of survivors of HNC, including dietary changes and weight loss in individuals who have inherently decreased oral intake secondary to their pathology, decreased social interactions, and an increased risk of depression and anxiety.^29,62-65 Pre-existing depression and depression diagnosed after HNC diagnosis are associated with a greater hazard of both cancer-specific and all-cause mortality.⁶⁶ Several robust prospective studies have demonstrated a link between mental disorders, including major depressive disorder, with increased risk of age-related medical illnesses, such as cardiovascular disease, type 2 diabetes, metabolic syndrome, and neurodegenerative diseases.^67-72 Although the association between psychiatric disorders and poor health outcomes is well-established, the association between sensorineural dysfunction, psychopathology, and cancer outcomes, including survival, is poorly understood. Assessment of olfaction in patients with cancer, particularly in at-risk individuals with HNC, represents an opportunity to further investigate these biopsychosocial associations.

Though there are many strengths of this unique investigation, the conclusions are limited by a cohort of participants with heterogenous disease and treatment types. The overarching association between OD and frailty is clear, yet clinical variations should be minimized in future studies to better understand how the present study鈥檚 findings apply to different pathologies, therapies, and outcomes. Acknowledging the relatively low disease stage in the surgical cohort and the lack of association between complications and frailty status in this current study, future studies should be specifically designed to have more balanced disease severity and be adequately powered to detect differences in postoperative complications. While this pilot, proof-of-concept study can provide valuable insights, additional limitations are inherent to a single-site study cohort, including limited generalizability to other populations, such as those with lower prevalence of frailty or from different geographic locations; lack of a control group; and selection bias from the lack of randomization.

Despite these limitations, there are multiple opportunities for future investigations. Prospective, longitudinal, multicenter studies are necessary to evaluate the outcomes associated with OD and the interaction with frailty status. Keeping in mind the important balance of brevity and clinical utility, the identification of an ideal olfactory assessment will be critical. For instance, the use of a shorter test, such as the Brief Smell Identification Test, which tracks well with the UPSIT, may offer similar accuracy.^73,74 From a mechanistic perspective, olfactory subdomains may offer insight into the association between outcomes and peripheral vs central sensorineural function.⁷⁵ Finally, it will be important to assess whether targeted olfactory interventions can make meaningful differences in clinical and oncologic outcomes. Olfactory testing holds potential to serve as a screening measure to either augment currently available frailty assessments or prompt risk-stratified and patient-specific frailty-associated interventions and rehabilitation.

Although patients with HNC are unaware of olfactory changes, disruption of this special sensorineural function is common and may serve as a bellwether of frailty. In this prospective cohort study, we demonstrated a dose-dependent association between increased levels of psychophysical OD and higher levels of frailty. These findings suggest that OD may function as an easily accessible, clinical measure for the assessment of frailty that potentially overcomes the shortcomings of current frailty assessments. Though validation is needed, the use of olfactory assessments in patients with HNC may represent a promising opportunity to identify at-risk individuals and develop interventions aimed at improving patient-centered care and oncologic outcomes.

Accepted for Publication: June 5, 2023.

Published Online: July 27, 2023. doi:10.1001/jamaoto.2023.1935

Corresponding Author: Nicholas R. Rowan, MD, Department of Otolaryngology鈥揌ead and Neck Surgery, The Johns Hopkins University School of Medicine, 601 N Caroline St, Sixth Floor, Baltimore, MD 21287 (nrowan1@jhmi.edu).

Author Contributions: Dr Rowan 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.

Concept and design: Mady, De Ravin, Vohra, Lu, Rowan.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Mady, De Ravin, Lu, Rowan.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Mady, De Ravin, Vohra, Dalton, Rowan.

Obtained funding: Mady, Rowan.

Administrative, technical, or material support: Mady, De Ravin, Vohra, Lu, Newman.

Supervision: Newman, Hall, Rowan.

Conflict of Interest Disclosures: Drs Mady and Rowan reported grants from the University of Pennsylvania Department of Otorhinolaryngology during the conduct of the study. Dr Hall reported grants from the National Institutes of Health (NIH), the Veterans Health Administration Office of Research and Development, and unpaid consulting with FutureAssure, LLC outside the submitted work. Dr Rowan is supported by the Johns Hopkins University Claude D. Pepper Older Americans Independence Center, funded by the National Institute on Aging and NIH (P30AG021334). No other disclosures were reported.

Data Sharing Statement: See Supplement 2.

Additional Contributions: We thank Justin R. Shinn, MD, Robert M. Brody, MD, Steven B. Cannady, MD, Karthik Rajasekaran, MD, and Rabie M. Shanti, MD, for participation in patient care and study enrollment efforts. None of these individuals were compensated.