Low Skeletal Muscle Mass as a Key Predictor of Postoperative Diet Outcomes in Circumferential Hypopharyngeal Cancer Reconstruction

ABSTRACT

Objective

To evaluate the association between preoperative lumbar skeletal muscle index (LSMI) and postoperative dietary outcomes (regular diet versus texture‐modified diet) in patients undergoing circumferential hypopharyngeal cancer resection with tubed anterolateral thigh flap reconstruction.

Methods

This single‐center retrospective cohort study included 73 patients between 2009 and 2020. LSMI was calculated from lumbar skeletal muscle cross‐sectional area at the C3 vertebral level on preoperative CT scans. The primary outcome was dietary restriction, defined as Functional Oral Intake Scale ≤ 6 at ≥ 12 months postoperatively. Candidate predictors (pre‐, intra‐ and postoperative) with p < 0.20 in univariate screening underwent theory‐driven confounder assessment based on directed acyclic graphs, prior literature, and clinical knowledge. Change‐in‐estimate criterion (Δβ ≥ 10%) supported selecting confounding. LSMI components (C3CSA, L3CSA, age, sex, weight, height, BMI) were excluded from adjustment to prevent overadjustment for mediators and multicollinearity. The final model analyzed LSMI as a continuous predictor with logistic regression.

Results

Fifteen patients (20.5%) developed dietary restriction (FOIS < 7). Three confounders (Eichner A, Stage IV, hemoglobin) for LSMI were identified. Multivariable adjustment improved model fit (AIC: 65.8 → 63.4; Nagelkerke R ²: 0.237 → 0.383) while maintaining statistical significance. Each 1 cm²/m² increase in LSMI was associated with a 16.1% reduction in the odds of FOIS < 7 (adjusted OR: 0.839; p = 0.006). The optimal LSMI cut‐off value was 42.82 cm²/m².

Conclusion

Preoperative LSMI independently predicts long‐term dietary outcomes after hypopharyngeal reconstruction. Restricting the cohort to uniform resection and reconstruction methods clarified the relationship between skeletal muscle mass and swallowing function, enabling precise risk stratification and supporting targeted prehabilitation interventions.

Level of Evidence

4.

This study found that preoperative skeletal muscle mass (measured by LSMI) strongly predicts long‐term dietary outcomes after complex hypopharyngeal cancer reconstruction. An LSMI cutoff value of 42.82 was established, revealing that patients with low muscle mass had a significantly 10‐fold higher risk of requiring a restricted or texture‐modified diet one year post‐surgery. These results suggest that assessing a patient's skeletal muscle mass before surgery is a valuable tool for predicting swallowing function recovery.

1. Introduction

Swallowing function is a primary outcome in head and neck cancer (HNC) treatment and has a significant impact on patients' postoperative quality of life [1, 2]. Several factors have been reported to influence postoperative swallowing function, including the size and location of the original tumor, the structures involved, and surgical/non‐surgical or reconstructive treatment options [1].

Recent studies have demonstrated a strong association between swallowing function and sarcopenia, a condition characterized by the loss of skeletal muscle mass (SMM) and function [3, 4, 5]. Sarcopenia has been shown to negatively affect swallowing muscles, leading to impaired swallowing ability in various clinical conditions [6]. This results in restrictions on dietary texture and a decline in patients' quality of life [7]. SMM is typically assessed using computed tomography (CT) at the third lumbar vertebra (L3) level. However, routine abdominal CT scans often do not extend to the L3 level as part of preoperative imaging for HNC. As an alternative, SMM evaluation at the third cervical vertebra (C3) level has been validated as a feasible method [8, 9, 10, 11].

There is growing evidence suggesting that low SMM influences postoperative outcomes in HNC patients [9, 10, 12, 13, 14, 15, 16]. As such, low SMM has been associated with increased risk of intraoperative blood transfusions [15] and postoperative complications [9, 11, 12, 13, 14, 17], as well as prolonged hospital stay [13]. Furthermore, low SMM is recognized as a significant risk factor for chemotherapy dose‐limiting toxicity (CDLT), which subsequently leads to reduced overall survival [18]. However, despite the well‐established relationship between sarcopenia and swallowing dysfunction, few studies have specifically examined the association between preoperative SMM and postoperative swallowing function in HNC patients [19]. Filling this evidence gap is crucial for determining whether preoperative SMM should be considered when evaluating postoperative swallowing function for the development of more accurate prognostic models and targeted prehabilitation strategies that improve postoperative outcomes [20].

This study therefore aims to investigate the relationship between pre‐operative SMM assessed from CT at C3 level and postoperative long‐term dietary outcomes in patients undergoing circumferential resection and reconstruction for hypopharyngeal cancer, a setting with limited variability in resection extent and reconstruction methods.

2. Materials and Methods

This study was conducted as a retrospective cohort study and was approved by the Ethics Committee of The Jikei University School of Medicine (Approval No. 33‐215(10832)).

2.1. Study Population

We included all patients who underwent primary tumor resection and tubed anterolateral thigh (ALT) flap reconstruction for hypopharyngeal cancer between September 2009 and September 2020. The inclusion criteria for the analyses were as follows: (1) recent CT scanning (within 2 months before surgery) and (2) assessment of dietary status at least 12 months postoperatively or after additional treatment. Patient data were collected from medical records (Figure 1).

FIGURE 1.

Patient selection flowchart.

2.2. Potential Confounders

Standard demographics, comorbidities, preoperative laboratory data, occlusal status (Eichner classification), tumor/treatment characteristics, and intra‐ and postoperative factors were evaluated as potential confounders (Table 1).

TABLE 1.

Univariate logistic regression analysis of all perioperative factors for predicting FOIS < 7.

		All patients (n = 73)	Last dietary status		p	Crude OR	95% CI
			FOIS = 7 (n = 58)	FOIS < 7 (n = 15)
Preoperative factors
Age		71 (63–77)	70.5 (61.5–74.8)	72 (69.5–80)	0.069	1.076	0.994–1.165
Sex	Male	63 (86.3)	52 (89.7)	11 (73.3)	0.114	3.152	0.760–13.072
	Female	10 (13.7)	6 (10.3)	4 (26.7)
Height (cm)		163 (159–167.8)	164 (159.3–168)	160 (156–166.6)	0.143	0.941	0.867–1.021
BW (kg)		55 (48–61)	56 (49–63.8)	50.4 (46.5–54)	0.082	0.948	0.892–1.007
BMI (kg/m2)		20.2 (18.5–22.5)	20.7 (18.7–22.7)	19.9 (18.3–20.8)	0.223	0.891	0.740–1.073
Past history	HT	25 (34.2)	20 (34.5)	5 (33.3)	0.933	0.950	0.285–3.162
	DM	11 (15.1)	7 (12.1)	4 (26.7)	0.170	2.649	0.659–10.643
	CAD	2 (2.7)	2 (3.4)	0 (0)	1.000	—	—
	CHF	1 (1.4)	1 (1.7)	0 (0)	1.000	—	—
	COPD	3 (4.1)	3 (5.2)	0 (0)	1.000	—	—
	PVD	2 (2.7)	1 (1.7)	1 (6.7)	0.331	4.071	0.240–69.185
	CKD	3 (4.1)	1 (1.7)	2 (13.3)	0.086	8.769	0.738–104.182
Prior RT		18 (24.7)	13 (23.2)	5 (33.3)	0.385	1.731	0.502–5.971
Prior CT		11 (15.1)	8 (13.8)	3 (20.0)	0.552	1.563	0.360–6.788
Stage	I	2 (2.7)	0 (0)	2 (13.3)	1.000	—	—
	II	3 (4.1)	2 (3.4)	1 (6.7)	0.582	2	0.169–23.665
	III	9 (12.3)	6 (10.3)	3 (20.0)	0.319	2.167	0.473–9.921
	IV	57 (78.1)	48 (82.8)	9 (60.0)	0.044	0.276	0.079–0.965
Lymph node metastasis		44 (60.3)	38 (65.5)	6 (40.0)	0.078	0.351	0.109–1.126
Previous abdominal surgery		11 (15.1)	7 (12.1)	4 (26.7)	0.170	2.649	0.660–10.643
Preoperative blood sample tests	ChE (U/L)	269 (204–295)	269 (205–302)	259 (194–293)	0.580	0.998	0.989–1.006
	TP (g/L)	7.0 (6.4–7.4)	7.1 (6.4–7.5)	6.9 (6.4–7.2)	0.166	0.497	0.185–1.335
	Alb (g/L)	3.8 (3.2–4.2)	3.7 (3.2–4.2)	3.9 (3.2–4.0)	0.908	0.944	0.354–2.515
	Cr (mg/dL)	0.7 (0.6–0.9)	0.8 (0.6–0.9)	0.7 (0.6–1.0)	0.087	1.941	0.909–4.146
	CRP (mg/L)	0.42 (0.11–1.53)	0.41 (0.11–1.53)	0.61 (0.13–1.74)	0.849	0.969	0.699–1.342
	WBC (109/L)	6.8 (5.4–8.1)	6.9 (5.5–8.2)	6.0 (5.3–7.8)	0.164	0.777	0.545–1.108
	Lymph (%)	20.7 (15.4–24.3)	20.9 (16.4–25.4)	20.2 (14.7–22.1)	0.467	0.973	0.903–1.048
	Neutro (%)	70.0 (64.4–76.0)	70.1 (64.3–76.0)	69.7 (66.2–76.3)	0.709	1.012	0.952–1.075
	Hb (g/L)	12.4 (10.8–13.7)	12.2 (10.8–13.6)	13.1 (11.0–13.8)	0.183	1.171	0.928–1.478
	Plt (109/L)	255 (210–328)	255 (211–336)	257 (211–326)	0.290	0.997	0.991–1.003
	NLR	3.39 (2.62–4.90)	3.40 (2.49–4.64)	3.39 (2.95–5.05)	0.783	1.029	0.839–1.264
	PNI	44.7 (38.7–48.5)	45.0 (38.7–48.2)	43.4 (37.9–48.8)	0.487	0.97	0.889–1.058
	GNRI	99.8 (87.3–107.2)	100.8 (87.6–107.8)	95.8 (87.1–103.6)	0.518	0.986	0.943–1.030
C3CSA		2666 (2362–3003)	2873 (2434–3108)	2392 (2094–2635)	0.006	0.998	0.997–0.999
L3CSA		75.2 (64.6–81.0)	77.4 (67.8–85.1)	64.3 (59.5–73.8)	0.017	0.943	0.898–0.990
LSMI (cm2/m2)		46.0 (40.7–49.5)	47.1 (42.3–50.9)	40.0 (38.2–42.4)	0.003	0.858	0.774–0.951
Eichner classification	A1	13 (17.8)	13 (22.4)	0 (0)	0.035	0.183	0.038–0.888
	A2	9 (12.3)	8 (13.8)	1 (6.7)
	A3	6 (8.2)	5 (8.6)	1 (6.7)
	B1	10 (13.7)	8 (13.8)	2 (13.3)	0.867	1.117	0.306–4.072
	B2	4 (5.5)	2 (3.4)	2 (13.3)
	B3	3 (4.1)	3 (5.2)	0 (0)
	B4	1 (1.4)	1 (1.7)	0 (0)
	C1	9 (12.3)	5 (8.6)	4 (26.7)	0.036	3.529	1.086–11.470
	C2	5 (6.8)	4 (6.9)	1 (6.7)
	C3	11 (15.1)	8 (13.8)	3 (20.0)
Intraoperative factors
Neck dissection	Non	9 (12.3)	6 (10.3)	3 (20.0)	0.319	2.167	0.473–9.921
	Hemi—	6 (8.2)	3 (51.7)	3 (20.0)	0.082	4.583	0.823–25.540
	Bi—	58 (79.5)	49 (84.5)	9 (60.0)	0.065	0.312	0.091–1.077
Intraoperative re‐anastomosis		6 (8.2)	5 (8.6)	1 (6.7)	0.794	0.743	0.08–6.885
Superior pharyngeal cut		14 (19.2)	10 (17.2)	4 (26.7)	0.412	1.745	0.461–6.612
Flap length 8 cm		16 (21.9)	13 (22.4)	3 (20.0)	0.816	0.846	0.207–3.460
Recipient artery	STA	41 (56.2)	30 (51.7)	11 (73.3)	0.141	2.567	0.732–9.003
	TCA	19 (26.0)	16 (27.6)	3 (20.0)	0.553	0.656	0.163–2.635
	Other	13 (17.8)	12 (20.7)	1 (6.7)	0.232	0.274	0.033–2.295
Recipient vein	IJV	71 (97.3)	56 (96.6)	15 (100.0)	1.000	—	—
	Other	2 (2.7)	2 (3.4)	0 (0)	1.000	—	—
Recipient vein ≧ 2		29 (39.7)	23 (39.7)	6 (40.0)	0.981	1.014	0.318–3.234
Operation time (min.)		484 (442–559)	490 (442–563)	481 (445–516)	0.352	0.997	0.991–1.003
Ischemia time (min.)		96 (85–116)	94 (85–116)	98 (91–120)	0.471	1.008	0.987–1.029
Blood Loss (mL)		180 (156–298)	178 (110–293)	210 (130–288)	0.521	1.001	0.998–1.005
Transfusion		10 (13.7)	7 (12.1)	3 (20.0)	0.431	1.821	0.410–8.094
Postoperative factors
Complication	General	3 (4.0)	1 (1.7)	2 (13.3)	0.086	8.769	0.738–104.182
	Neck	14 (19.2)	10 (17.2)	4 (26.7)	0.412	1.745	0.461–6.612
	Flap donor	13 (17.8)	11 (19.0)	2 (13.3)	0.6133	0.657	0.129–3.345
	Flap	2 (2.7)	2 (3.4)	0 (0)	1.000	—	—
Neck fistula		5 (6.8)	5 (8.6)	0 (0)	0.808	39,328	0‐inf
Reoperation		3 (4.1)	2 (3.4)	1 (6.7)	0.742	1.333	0.241–7.385
Nasal reflux		16 (21.9)	13 (22.4)	3 (20.0)	0.792	0.827	0.202–3.384
Endoscopic dilation		12 (16.4)	7 (12.1)	5 (33.3)	0.061	3.571	0.941–13.55
Postoperative RT		26 (35.6)	22 (37.9)	4 (26.7)	0.420	0.595	0.169–2.101

Note: Number of patients (%) or Median (Q1–Q3). Bold indicates the significant values related to LSMI.

Abbreviations: BW: body weight; C3CSA: cross‐sectional area at C3 level; CAD: chronic arterial disease; CHF: chronic heart failure; CKD: chronic kidney disease; COPD: chronic obstructive pulmonary disease; CT: chemotherapy; DM: diabetes mellitus; GNRI: geriatric nutritional risk index = 1.487 × ALB (g/L) + 41.7 × present BW/ideal BW (kg); HT: hypertension; IJV: internal jugular vein; L3CSA: cross‐sectional area at L3 level; LSMI: lumber skeletal muscular index; NLR: neutrophil‐to‐lymphocyte ratio; PNI: prognostic nutritional index = 10 × Alb + 0.05 × total lymphocyte count; PVD: peripheral vascular disease; RT: radiation therapy; STA: superior thyroid artery; TCA: transcervical artery.

2.3. Surgical Technique

The surgical technique followed the approach described by Yu et al. [21]. The ALT free flap–oral/esophageal anastomosis was performed circumferentially, with a flap width of either 8 or 10 cm on the esophageal side [22] (Figure 2A–C). The surgeon determined the width of the ALT free flap based on the patient's body shape or the width of the esophagectomy edge. On the cervical esophageal side, spatulation was performed on the posterior wall, and a small triangle of the ALT free flap, predesigned with a 15‐mm edge, was inserted in that portion. The fascia lata of the ALT free flap was wrapped around the neopharynx to reinforce the suture lines and prevent fistula formation at the anastomosis. The practice of performing dual venous anastomosis was introduced at a certain period based on previous literature [23].

FIGURE 2.

(A) Flap design, (B) tuberalization, (C) concept design illustrations.

At our institution, a routine barium swallow study is performed on postoperative day 14 before resuming oral intake, resulting in most patients beginning oral intake on the 14th day. Complications, cervical fistula formation, and reoperations were counted as early postoperative complications occurring within 30 days.

2.4. Exposure

SMM was assessed using pre‐treatment CT scans of the head and neck region. The cross‐sectional area (CSA) at the level of the third cervical vertebra (C3), including both the transverse processes and the entire vertebral arch, was segmented using Hounsfield unit thresholds of −29 to +150. In cases where the sternocleidomastoid muscle (SCM) was invaded by lymph node metastasis, the area of the contralateral SCM was doubled for calculation. The muscle CSA at C3 (C3CSA) was measured using ImageJ v1.52a (Wayne Rasband, National Institutes of Health, Bethesda, MD) (Figure 3). Based on previous reports [8, 9, 24, 25], the lumbar skeletal muscle index (LSMI) was calculated by adjusting C3CSA for age, sex, height, and weight (Figure 3).

FIGURE 3.

Measurement of C3CSA using ImageJ, and formula for calculating Lumber SMI.

2.5. Outcomes

The primary outcome of this study was dietary status 12 months after surgery, assessed through patient interviews during the scheduled follow‐ups in the clinic. Postoperative dietary status was evaluated using the Functional Oral Intake Scale (FOIS) [26], a widely recognized tool for measuring oropharyngeal dysphagia. The FOIS assigns scores ranging from 1 (no oral intake) to 7 (total oral intake with no restrictions), with higher scores indicating improved swallowing function. The scores of 1 to 3 were defined as tube‐dependent, while scores of 4 to 6 indicated total oral intake but with specific restrictions. A score of 7 specified a total oral diet with no restrictions. For the purpose of this study, a score of 7 on the FOIS was classified as a regular diet group, while all other scores were categorized as a texture‐modified diet group.

2.6. Statistical Analyses

All analyses were performed using Python 3.9 (statsmodels, scikit‐learn). The primary outcome was postoperative diet restriction, defined as FOIS ≤ 6 at ≥ 12 months.

Univariate screening identified candidate predictors with p < 0.20, including LSMI. Confounder assessment followed a theory‐driven approach integrating directed acyclic graphs (DAG), prior literature (tumor stage [27], oral function [27, 28] and nutritional status [29]), and clinical knowledge. DAG defined as X ← Z → Y, where X represents the primary exposure (LSMI), Y denotes the primary outcome (FOIS < 7), and Z represents a set of potential confounding factors, including tumor stage, oral function, and nutritional status. The change‐in‐estimate criterion (|Δβ| ≥ 10%) was applied to supplementally confirm confounding. LSMI components (C3CSA, L3CSA, age, sex, weight, height, BMI) were excluded from adjustment to prevent overadjustment for mediators/components and multicollinearity [30].

Confounding was assessed by comparing crude and adjusted models. Individual adjustment for each confounder and multivariable adjustment were performed using logistic regression. Nonlinearity was evaluated using restricted cubic splines (three knots) with likelihood‐ratio testing. An optimal cutoff was derived via Youden index for clinical application, though continuous LSMI was prioritized for risk estimation to preserve statistical power.

3. Results

A total of 86 patients underwent primary tumor resection and tubed ALT flap reconstruction for hypopharyngeal cancer between September 2009 and September 2020. Among these, 73 met the criteria to be included in the study cohort. Median age was 71, and 86.3% were male. At postoperative 12 months, 79.5% (n = 58) returned to regular diet, while 20.5% (n = 15) was only able to ingest a texture‐modified diet. Table 1 presents a comparison of preoperative, intraoperative, and postoperative factors between the regular diet versus texture‐modified diet groups.

Regarding preoperative factors, significant differences were observed between the two groups in C3CSA, L3CSA, and LSMI. The median LSMI was 48.4 cm²/m² in the regular diet group compared with 40.0 cm²/m² in the texture‐modified diet group (p < 0.001). Regarding intraoperative factors, no significant differences were found between the two groups for any intraoperative factors. About 14 patients required a superior pharyngeal cut: 7 involving the unilateral lateral wall, 5 involving the posterior wall, 1 involving bilateral lateral walls with the posterior wall, and 1 involving the tongue base.

Regarding postoperative factors, no significant differences were observed between the two groups. General complications included life‐threatening arrhythmia, congestive heart failure, and pneumonia, all of which were managed conservatively. Cervical complications consisted exclusively of wound dehiscence, which was also managed conservatively. Flap donor site complications were all cases of delayed wound healing due to skin graft failure. Reoperations were performed for flap total loss and cervical complications. Regarding final dietary function, no patients remained dependent on tube feeding (although two patients temporarily required tube feeding; their conditions improved following endoscopic and surgical interventions). According to the FOIS [26], seven patients were classified as Level 4 (Total oral diet of a single consistency), six as Level 5 (Total oral diet with multiple consistencies, but requiring special preparation or compensations), and two as Level 6 (Total oral diet with multiple consistencies without special preparation, but with specific food limitations).

Among all examined variables, significant differences between the two groups were observed only in C3CSA, L3CSA, and LSMI. LSMI was selected as the primary predictor because it better reflects whole‐body muscle mass, permits clearer clinical interpretation, and is a standard sarcopenia indicator.

Theory‐driven confounder selection (DAG, literature, clinical knowledge) identified Stage IV, Eichner classification, and hemoglobin (Hb) as confounders (Δβ ≥ 10%; Table 2). LSMI components (C3CSA, L3CSA, age, sex, weight, height, BMI) were excluded to avoid overadjustment.

TABLE 2.

Assessment of confounding effects on LSMI based on clinical domains.

Variable	OR	95% CI		p	Δβ (%)	Rationale
		Lower	Upper
Lumber SMI	0.858	0.774	0.951	0.003	—	Main factor
C3CSA	0.998	0.997	0.999	0.006	—	Mediator/component
L3CSA	0.943	0.898	0.990	0.017	—	Mediator/component
Eichner A	0.183	0.038	0.888	0.035	−15.0	Oral function
Eichner C	3.529	1.086	11.470	0.036	−9.2	Oral function
Endoscopic expansion	3.571	0.941	13.551	0.061	4.5	Postoperative
Bi‐neck dissection	0.313	0.091	1.077	0.065	9.4	Other
Stage IV	0.313	0.091	1.077	0.065	18.5	Stage
Age	1.077	0.994	1.165	0.069	—	Mediator/component
Lymph node metastasis	0.351	0.109	1.126	0.078	0.8	Other
Body weight	0.948	0.892	1.007	0.082	—	Mediator/component
Hemi‐neck dissection	4.583	0.823	25.541	0.082	5.3	Other
General complication	8.769	0.738	104.187	0.086	−3.9	Postoperative
CKD	8.769	0.738	104.187	0.086	3.7	Other
Cr	1.941	0.909	4.146	0.087	1.3	Other
Sex	3.152	0.760	13.072	0.114	—	Mediator/component
STA	2.567	0.732	9.003	0.141	0.8	Other
Height	0.941	0.868	1.021	0.143	—	Mediator/component
WBC	0.777	0.545	1.108	0.164	1.8	Other
TP	0.497	0.185	1.335	0.166	−3.6	Nutritional Status
Prev_Abd_Sx	2.649	0.660	10.643	0.170	−3.0	Other
DM	2.649	0.660	10.643	0.170	−2.8	Other
Hb	1.171	0.928	1.478	0.183	13.5	Nutritional Status

Note: Candidate variables identified through univariate screening (p < 0.20) were assigned to pre‐specified confounder sets (tumor stage, oral function, and nutritional status) to assess their impact on the LSMI regression coefficient (Δβ).

Abbreviations: C3CSA: cross‐sectional area at C3 level; CI: confidence interval; CKD: chronic kidney disease; Cr: creatinine; DM: diabetes mellitus; Hb: hemoglobin; L3CSA: cross‐sectional area at L3 level; LSMI: lumber skeletal muscular index; OR: odds ratio; Prev_Abd_Sx: previous abdominal surgery; STA: superior thyroid artery (recipient artery); TP: total protein; WBC: white blood cell.

The adjusted model treating LSMI as a continuous variable demonstrated superior goodness‐of‐fit compared to the unadjusted model, as evidenced by a reduction in AIC and an increase in Nagelkerke R ². Furthermore, an additional model utilizing dichotomized LSMI—based on the optimal cutoff derived from ROC analysis—consistently showed that low LSMI was associated with an increased risk of FOIS < 7 after adjustment, confirming that low LSMI serves as a clinically identifiable risk marker (Table 3). The predicted probability curve demonstrated a continuous relationship between LSMI and FOIS < 7, showing a monotonic decrease in risk with increasing LSMI. For each 1 cm²/m² increase in LSMI, the odds of dietary restriction decreased by 16.1% (OR = 0.839) based on the adjusted model (Figure 4). An LSMI cutoff of 42.81 cm²/m² achieved 80.0% sensitivity and 72.4% specificity (Figure 5).

TABLE 3.

Sensitivity analysis for the association between LSMI and postoperative diet restriction: Assessment of confounding effects.

Model		OR	95% CI	p (Wald's)	p (LR)	AIC	AUC	Log‐likelihood	R 2 (Nagelkerke)
Continuous LSMI	Crude	0.858	0.775–0.951	0.003	0.0006	65.8	0.808	−30.919	0.237
	Adjusted	0.839	0.742–0.950	0.006	0.0004	63.4	0.828	−26.719	0.383
Binary LSMI	Crude	11.20	2.77–45.23	< 0.001	0.0001	63.1	—	−29.570	0.286
	Adjusted	13.77	2.76–68.65	0.001	0.0002	61.1	—	−25.544	0.421

Note: No. of observations = 72 (one missing for Eichner A); binary: < 42.81 versus ≧ 42.81; adjusted: stage IV (binary), Eichner A (binary), Hb (continuous); Hosmer–Lemeshow χ ²: 5.330; Hosmer–Lemeshow p: 0.722.

FIGURE 4.

Predicted probability curve for diet restriction (FOIS < 7).

FIGURE 5.

(A) Distribution of LSMI values in the cohort, (B) ROC curve for LSMI predicting diet restriction.

4. Discussion

Swallowing function is closely related to quality of life and is a crucial factor in evaluating outcomes of pharyngo‐laryngectomy. Various definitions of dysphagia have been adopted in research; some studies define it as the inability to consume a normal diet—specifically, the necessity for adapting solid foods for ingestion [31, 32]. The FOIS Level 7, representing “total oral diet with no restrictions,” provides a clinically validated cutoff that distinguishes normal swallowing function from any dietary modification (FOIS < 7) [33, 34]. Long‐term restrictions in dietary texture have been associated with declines in both psychological and physical aspects of quality of life [35], reduced nutritional intake [35, 36], and difficulties in maintaining oral muscle strength, which may contribute to the progression of sarcopenia [37]. Furthermore, a reduction in chewing frequency due to modified diet textures has been suggested to be associated with cognitive function [38], oral function [39], and appetite for eating [39].

HNC is presumed to be associated with sarcopenia because the affected anatomical sites are involved in nutritional intake. Multiple studies have reported on the relationship between preoperative sarcopenia and postoperative outcomes, which has been systematically reviewed by Ansari et al. [16]. However, this review did not include findings related to postoperative swallowing function. A possible reason is the difficulty in maintaining a homogeneous patient cohort in HNC studies, as resection extent varies between cases, and factors such as nerve paralysis and rehabilitation further influence swallowing function. In contrast, this study exclusively focused on patients who underwent total pharyngolaryngectomy with tubed ALT flap reconstruction, as this group presents relatively less variation in resection extent compared to other surgical procedures, ensuring a higher degree of cohort homogeneity.

The original definition of sarcopenia is based on lumbar SMM. However, imaging in HNC patients is often limited to cervical CT (for primary tumor evaluation) and chest CT (to evaluate pulmonary metastases). Therefore, alternative methods using SMM at the C3 vertebral level have been explored. This approach has allowed the use of C3‐derived sarcopenia as an evaluation criterion in many HNC related studies [8, 9, 10, 11]. However, debates remain: some question the utility of CT‐based SMM alone without functional measures [40], others emphasize muscle quality (myosteatosis) over quantity [41], and the necessity of converting C3CSA to lumbar equivalents has been challenged [17]. It is important to note that LSMI alone is insufficient to diagnose sarcopenia. Sarcopenia requires both low SMM and reduced muscle strength, with evaluations typically including grip strength, walking speed, or the ability to rise from a chair [42]. Furthermore, while various software tools [14, 15, 43, 44, 45, 46] have been employed for measurement, there is no standardized method for measuring muscle CSA across studies. In this study, ImageJ was utilized, although potential biases due to differences in measurement methods across institutions cannot be ruled out. This study focused solely on evaluating the relationship between low SMM calculated by cervical CT images and swallowing function.

The median LSMI in the present cohort was 46 cm²/m², slightly higher than typical values reported in existing literature [47, 48]. Although several studies have utilized specific cutoff points for sarcopenia [14, 15], dichotomizing continuous variables is generally discouraged as it leads to a loss of information and reduced statistical power. Consequently, we analyzed LSMI as a continuous variable to maintain the granularity of the data [49]. We concluded that the adjusted model was superior to the unadjusted model for causal inference, as demonstrated by improved goodness‐of‐fit (AIC) and explanatory power (Nagelkerke R ²). This finding suggests that preoperative tumor stage, Eichner classification, and overall nutritional status are indirectly associated with postoperative dietary outcomes. However, the selection of Hb as a key confounder—rather than other established nutritional markers such as serum albumin [50], white blood cell count [50], neutrophil‐to‐lymphocyte ratio (NLR) [51], or the prognostic nutritional index (PNI) [52]—remains a point of discussion.

Previous studies have examined the relationship between residual mucosa and swallowing function after laryngectomy [53], but few studies have explored neopharyngeal biomechanics during swallowing. Early manometric research showed that laryngectomized patients exhibit reduced pharyngeal contraction pressure and prolonged bolus transit times due to pharyngeal neuromuscular dysfunction [54, 55]. Other reported causes of dysphagia following laryngectomy include decreased pharyngeal propulsion force, increased bolus pressure in the hypopharynx, pharyngeal muscle weakness, pharyngoesophageal spasms, pharyngoesophageal or neopharyngeal stricture, and nasopharyngeal reflux [50, 56]. In this study, no association was found between superior pharyngeal cut and dietary outcomes, suggesting that further accumulation of cases is needed to evaluate this relationship. Based on our clinical experience, porous foods like soba noodles tend to adhere to the pharyngeal wall, often causing neopharyngeal obstruction due to food residue. A reduction in LSMI might relate to pharyngeal muscle weakness; however, further investigations such as videofluoroscopic swallowing study (VFSS) [57] and the DIGEST rating scale (efficiency and safety modules) [58] are needed to confirm this relationship.

This study found that low LSMI was associated with a tenfold higher relative risk for requiring a texture‐modified diet, suggesting the possibility of predicting decline in quality of life related to postoperative dietary restrictions preoperatively. Preoperative LSMI measurements could therefore help identify patients who would benefit from proactive perioperative rehabilitation, including prehabilitation. To widely utilize this knowledge, an efficient tool for LSMI measurement is essential. Specifically, a software solution should be developed to read CT DICOM data, identify the C3 slice, segment the paraspinal and SCM muscles, set the appropriate Hounsfield window, and calculate the CSA. Ultimately, LSMI should be calculated automatically by uploading an anonymized neck axial scan DICOM file. Evaluating preoperative SMM and sarcopenia is increasingly recommended [59, 60] and our findings offer valuable insights into personalized perioperative interventions.

While HNC prehabilitation is reported for radiotherapy and chemotherapy [61, 62], few studies focus on surgery. A potential limitation of surgical prehabilitation is the brief diagnosis‐to‐surgery interval, which may restrict its effectiveness. However, early postoperative exercise interventions may influence muscle mass more significantly than preoperative interventions [63], and oral motor training improves postoperative swallowing function after oral cancer surgery [64]. Therefore, future studies should investigate early postoperative exercise interventions for patients with low SMM undergoing HNC surgery.

This study effectively clarified the relationship between preoperative SMM and postoperative dietary outcomes by selecting patients with consistent surgical methods, minimizing typical procedural variability in HNC surgery. However, FOIS measures functional intake but lacks data on pathophysiology and pharyngeal efficiency. As a clinician‐reported scale, it also fails to capture patient‐reported outcomes (PROMs) or quality of life [65]; integrating DIGEST rating scale [58], MD Anderson Dysphagia Inventory (MDADI) [66], and the Eating Assessment Tool (EAT‐10) [67] would provide a more comprehensive evaluation. Furthermore, focusing on fasciocutaneous flaps and a single‐institution, homogeneous population limits generalizability to diverse populations or jejunal reconstruction. Finally, LSMI was estimated from C3‐level CT images. While C3‐to‐L3 conversion formulas are validated [8, 9, 24, 25], some suggest C3 measurements alone suffice [11]. Direct L3‐level data—the gold standard—were unavailable because lumbar CT is not routine in head and neck oncology. Future research should incorporate direct L3 measurements to further clarify the relationship between muscle mass and dietary outcomes.

5. Conclusion

Preoperative LSMI independently predicts long‐term dietary outcomes (FOIS) after hypopharyngeal reconstruction with free ALT flap. Each 1 cm²/m² increase in LSMI reduced the risk of dietary restriction by 16.1%, demonstrating a linear relationship. By restricting the cohort to uniform resection and reconstruction methods, this study effectively isolated the independent contribution of SMM to swallowing function. LSMI outperformed dental status and nutritional markers as a predictor, suggesting that whole‐body muscle mass reflects systemic functional reserve. Long‐term dietary restrictions lead to reduced quality of life and malnutrition; therefore, preoperative risk stratification is clinically important. The derived LSMI cutoff of 42.82 cm²/m² may guide patient counseling and initiation of prehabilitation interventions, though continuous LSMI is recommended for precise risk estimation. Further validation in multi‐institutional, ethnically diverse cohorts is necessary to confirm these findings and establish generalized thresholds.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.