Preoperatively Diagnosed Versus Undiagnosed Major Salivary Gland Cancer: Clinicopathological Characteristics, Treatment Strategies, and Oncological Outcomes

Abstract

Objective

To evaluate differences in clinicopathological characteristics, treatment strategies, and oncological outcomes between preoperatively diagnosed and undiagnosed major salivary gland cancers (MSGCs).

Study Design

Retrospective cohort study.

Setting

A single tertiary academic hospital.

Methods

Surgically confirmed MSGCs between 2012 and 2023 were included. The cohort was divided into the diagnosed and undiagnosed groups based on the preoperative diagnosis of MSGC determined by fine‐needle aspiration cytology (FNAC) or core‐needle biopsy. Clinicopathologic features, surgical strategies, and loco‐regional control (LRC) were compared between the groups.

Results

A total of 122 patients were enrolled. Among those, 49 (40.2%) were diagnosed preoperatively, while 73 (59.8%) were undiagnosed. The undiagnosed group had a higher reliance on FNAC (93.2% vs 51.0%, P < .001) and higher proportions of low/intermediate‐grade tumors (75% vs 26.5%, P < .001) and mucoepidermoid carcinoma (45.2% vs 18.4%, P < .001). These were verified as independent risk factors for preoperative undiagnosis in multivariate logistic regression analysis. The undiagnosed group was associated with delayed (60 vs 24 days, P < .001) and more limited surgical procedures, including lower frequency of near‐total/total parotidectomy (23.3% vs 42.9%, P = .001) and elective neck dissection (26.5% vs 61.6%, P < .001), compared to the diagnosed group. Grade‐ and stage‐matched analysis showed a trend towards worse LRC in the high‐grade or advanced‐stage undiagnosed MSGCs.

Conclusions

Preoperatively undiagnosed MSGCs exhibit distinct clinicopathological characteristics, treatment strategies, and oncological outcomes compared to preoperatively diagnosed cases, highlighting the critical role of preoperative diagnosis in the management MSGCs.

Accurate preoperative diagnosis is essential for establishing appropriate treatment strategies and providing preoperative counseling, particularly in malignant tumors. ¹ However, preoperative diagnosis of major salivary gland cancer (MSGC) remains challenging, with diagnostic errors reported in 30% to 70% of cases, which is substantially higher than in other head and neck cancers (HNCs). ² , ³ , ⁴ , ⁵ This challenge is largely attributable to the marked histopathological heterogeneity of salivary gland tumors and the rarity of MSGC, which accounts for only 1% to 5% of all HNCs. ¹ , ² , ⁶ , ⁷

Fine‐needle aspiration cytology (FNAC) is the most commonly used diagnostic modality for the preoperative evaluation of salivary gland tumors. ¹ , ⁶ , ⁸ However, its diagnostic performance is variable, with reported sensitivities ranging from 38% to 86% and nondiagnostic rates of up to 30%. ³ , ⁴ , ⁹ , ¹⁰ Core‐needle biopsy (CNB) has therefore been proposed as an alternative, demonstrating lower nondiagnostic rates and higher diagnostic accuracy than FNAC. ¹¹ , ¹² , ¹³ Despite these modalities, a substantial proportion of MSGCs remain undiagnosed preoperatively and are identified as malignant only after surgery, often performed under the assumption of benign disease. ² , ¹⁴ , ¹⁵ , ¹⁶ A study of 9672 patients undergoing parotidectomy for parotid tumors reported false‐negative rates of preoperative diagnosis of up to 57% for myoepithelial carcinoma and mucoepidermoid carcinoma (MEC). ⁶ In such cases, initial surgical management may not conform to guideline‐recommended treatment, potentially compromising optimal treatment strategies and oncologic outcomes, particularly in high‐grade or advanced‐stage MSGCs. ¹ , ⁸ , ¹⁷

Against this background, this study aimed to characterize the clinicopathologic features of MSGCs that were not diagnosed preoperatively by FNAC or CNB and to evaluate the impact of preoperative undiagnosis on treatment strategies and oncologic outcomes of MSGCs.

Methods

Patients

This study was conducted as a retrospective cohort study. The study protocol was approved by the institutional review board of Kyungpook National University Hospital (KNUH‐202410014), and the need for informed consent was waived in view of the retrospective nature of the study.

For study eligibility, patients with surgically confirmed primary MSGC between March 2012 and December 2023 were retrospectively identified. Inclusion criteria were: (1) upfront surgical treatment for primary MSGC, (2) preoperative pathological evaluation with FNAC or CNB within 1 year before surgery, and (3) a minimum follow‐up period of 12 months. Exclusion criteria included: (1) metastatic tumors from nonsalivary primary sites, (2) nonepithelial salivary malignancies (eg, lymphoma or sarcoma), (3) absence of preoperative FNAC or CNB, and (4) FNAC or CNB performed more than 1 year before surgery.

Determination of Preoperative FNAC and CNB

All FNAC and CNB procedures were performed under ultrasound guidance by experienced head and neck surgeons. The choice between FNAC and CNB was made a priori through shared decision‐making, based on operator judgment and patient preference, considering patient‐related factors (comorbidities, bleeding risk, and economic or insurance status) and tumor‐related characteristics (location, cystic change, calcification, vascularity, and accessibility).

FNAC was generally preferred (1) in patients at increased risk of postprocedural bleeding (eg, coagulopathy, antithrombotic use, high vascularity, or involvement of major vessels), (2) in tumors with cystic change, (3) when cytopathologic evaluation was considered sufficient for diagnosis, and/or (4) when a less invasive and lower‐cost approach was prioritized. In contrast, CNB was favored (1) in patients without increased risk of bleeding, (2) in tumors with calcifications, (3) when histopathologic evaluation was required to differentiate salivary gland tumors from lymphoproliferative disorders based on the imaging findings, and/or (4) when higher procedural cost was acceptable.

Pathological assessment of both preoperative FNAC/CNB specimens and postoperative surgical specimens was performed by a group of 3 experienced pathologists with expertise in head and neck pathology. Final histopathological diagnoses were rendered according to routine institutional practice, with consensus review in cases of diagnostic uncertainty.

Definition of Preoperative Diagnosis

The results of FNAC or CNB were categorized as nondiagnostic, benign, atypia of undetermined significance (AUS), or (suspicious) malignancy. Preoperative diagnosis of MSGC was defined as FNAC or CNB results indicating (suspicious) malignancy. All patients underwent CT and MRI preoperatively. If FNAC/CNB suggested malignancy, additional studies such as PET, chest CT, or bone scintigraphy were performed to further characterize the primary tumor and evaluate the presence of metastatic disease.

Assessment Parameters and Statistical Analysis

Patients were divided into diagnosed and undiagnosed groups according to the preoperative diagnosis of MSGC based on FNAC or CNB results. Primary outcome variables included: (1) preoperative and postoperative pathological characteristics, (2) treatment strategies, and (3) oncological outcomes.

Pathological assessment included preoperative FNAC/CNB results and postoperative final diagnoses with detailed histological characteristics. Treatment strategies were evaluated by the interval from preoperative diagnosis to surgery and the extent of surgery for the primary tumor and neck. For MSGCs of the parotid gland, the surgical extent was classified as partial, near‐total, or total parotidectomy. Partial parotidectomy involved limited tumor excision with an approximately 0.5 to 1 cm margin while preserving uninvolved parotid parenchyma. Total parotidectomy entailed complete removal of the superficial and deep lobes, whereas near‐total parotidectomy consisted of extensive gland resection with more conservative dissection around critical structures, including the facial nerve and external carotid artery.

Loco‐regional control (LRC) was defined as the primary oncological outcome, as it reflects the relevance of the surgical strategy employed as the primary treatment. LRC was analyzed using the Kaplan‐Meier method with log‐rank tests. To minimize confounding, a matching model incorporating preoperative diagnosis with histological grade or disease stage was employed. Disease staging followed the AJCC 8th edition; cases diagnosed earlier were retrospectively restaged using available TNM data. For timing analyses, time zero was defined as the date on which FNAC or CNB results were reported and the patient was classified as preoperatively diagnosed or undiagnosed.

Multivariate logistic regression analysis was performed to evaluate the association between clinicopathological characteristics and the risk of preoperative undiagnosis of MSGC. Additionally, multivariate Cox regression analysis was conducted to access potential risk factors for treatment failure in MSGC. The results of these regression analyses were reported as odds ratios (ORs) or hazard ratios (HRs), accompanied by 95% confidence intervals (CIs) and P‐values.

Statistical analyses were performed using SPSS version 22.0. Continuous variables are reported as median (interquartile range [IQR]) and compared using the Mann‐Whitney U test, while categorical variables were compared using the chi‐square or Fisher's exact tests. Statistical significance was defined as P < .05.

Results

Baseline Patient Characteristics

A total of 122 patients were included; 73 (59.8%) were preoperatively undiagnosed and 49 (40.2%) were preoperatively diagnosed with MSGC (Table 1). The undiagnosed group had a significantly higher proportion of females (47.9% vs 24.5%, P = .009) and never‐smokers (71.2% vs 46.9%, P = .016). FNAC was more frequently used in the undiagnosed group than in the diagnosed group (93.2% vs 51.0%, P < .001). The parotid gland was more frequently the primary site in the undiagnosed group, with borderline statistical significance (83.6% vs 69.4%, P = .064).

Table 1.

Baseline Patient Characteristics

	Diagnosed (n = 49)	Undiagnosed (n = 73)	P‐value
Sex
Male	37 (75.5%)	38 (52.1%)	.009
Female	12 (24.5%)	35 (47.9%)
Age (years)	60 (IQR, 20; range, 27‐88)	55 (IQR, 20; range, 18‐89)	.024
Smoking status
Never	23 (46.9%)	52 (71.2%)	.016
Ex	11 (22.5%)	6 (8.2%)
Current	15 (30.6%)	15 (20.5%)
Alcohol intake
None	31 (63.3%)	49 (67.1%)	.856
Social	15 (30.6%)	19 (26.0%)
Heavy	3 (6.1%)	5 (10.2%)
Charlson comorbidity index
0‐2	30 (61.2%)	52 (71.2%)	.248
≥3	19 (38.8%)	21 (28.8%)
Primary sites
Parotid gland	34 (69.4%)	61 (83.6%)	.064
Submandibular gland	15 (30.6%)	12 (16.4%)
Clinical presentation (multiple responses available)
None (incidentally found)	0 (0.0%)	3 (4.1%)	.362
Painless neck mass	43 (87.8%)	62 (84.9%)
Painful neck mass	5 (10.2%)	7 (9.6%)
Lymph node enlargement	1 (2.0%)	0 (0.0%)
Facial nerve palsy	2 (4.1%)	1 (1.4%)
Maximal tumor diameter on imaging study (cm)	2.5 (IQR, 1.1; range, 0.8‐5.8)	2.4 (IQR, 1.8; range, 0.9‐6.3)	.794
Preoperative diagnostic methods
FNAC	25 (51.0%)	68 (93.2%)	<.001
CNB	24 (49.0%)	5 (6.8%)

Abbreviations: CNB, core‐needle biopsy; FNAC, fine‐needle aspiration cytology; IQR, interquartile range.

Pattern of Misdiagnosis and Final Histological Results

Table 2 summarizes the patterns of misdiagnosis and the final histological results of the study cohort. Among the 73 patients in the undiagnosed group, the preoperative FNAC or CNB results were categorized as nondiagnostic in 10 (13.7%), benign in 35 (47.9%), and AUS in 28 (38.4%) patients.

Table 2.

Pattern of Preoperative Misdiagnosis and Final Histology

	Diagnosed (n = 49)	Undiagnosed (n = 73)	P‐value
Results of FNAC or CNB
Nondiagnostic	‐	10 (13.7%)	‐
Benign	‐	35 (47.9%)
Atypia with undetermined significance	‐	28 (38.4%)
Final histological diagnosis
Grade
Low	8 (16.3%)	39 (53.4%)	<.001
Intermediate	5 (10.2%)	16 (21.9%)
High	36 (73.5%)	18 (24.7%)
Subtype
Mucoepidermoid carcinoma	9 (18.4%)	33 (45.2%)	<.001
Adenoid cystic carcinoma	4 (8.1%)	11 (15.1%)
Myoepithelial carcinoma	2 (4.1%)	8 (11.0%)
Adenocarcinoma, not otherwise specified	6 (12.2%)	6 (8.2%)
Salivary duct carcinoma	25 (51.0%)	6 (8.2%)
Squamous cell carcinoma	1 (2.0%)	1 (1.4%)
Other (acinic cell carcinoma, basal cell adenocarcinoma, secretory cell carcinoma, etc.)	2 (4.1%)	8 (11.0%)

Abbreviations: CNB, core‐needle biopsy; FNAC, fine‐needle aspiration cytology.

The final histological grade of MSGC in the undiagnosed group was classified as low, intermediate, and high in 39 (53.4%), 16 (21.9%), and 18 (24.7%) patients, respectively. In contrast, the diagnosed group showed low‐, intermediate‐, and high‐grade disease in 8 (16.3%), 5 (10.2%), and 36 (73.5%) patients, respectively (P < .001).

The distribution of MSGC subtypes also differed significantly between the 2 groups (P < .001). MEC was the most common subtype in the undiagnosed group (33/73, 45.2%), whereas salivary duct carcinoma (SDC) predominated in the diagnosed group (25/49, 51.0%).

Clinicopathological Characteristics Associated With Preoperative Undiagnosis of MSGC

Table 3 summarizes the results of the multivariate logistic regression analysis for clinicopathological characteristics associated with the preoperative undiagnosis of MSGC. The use of FNAC, compared with CNB, as a preoperative diagnostic method (OR: 12.86; 95% CI: 3.57‐46.33; P < .001), low/intermediate‐grade tumors (OR: 3.29; 95% CI: 1.08‐12.10; P = .045), and the MEC subtype (OR: 4.80; 95% CI: 1.20‐19.15; P = .06) were identified as independent characteristics associated with an increased risk of preoperative undiagnosis of MSGC.

Table 3.

Multivariate Analysis of Clinicopathological Characteristics Associated With Preoperative Undiagnosis of Major Salivary Gland Cancer

Covariates	Odds ratio	95% confidence interval	P‐value
Sex
Male (reference)	1	‐	‐
Female	3.52	0.89‐13.97	.074
Age
>65 years (reference)	1	‐	‐
≤65 years	2.59	0.71‐9.45	.151
Smoking history
Yes (reference)	1	‐	‐
No	3.940	0.95‐16.33	.059
Alcohol history
Yes (references)	1	‐	‐
No	0.29	0.07‐1.23	.094
Charlson comorbidity index
≥3 (references)	1	‐	‐
1‐2	0.90	0.22‐3.64	.887
Primary site
Submandibular/sublingual (references)	1	‐	‐
Parotid	1.710	0.56‐5.25	.350
Preoperative diagnostic method
CNB (reference)	1	‐	‐
FNAC	12.86	3.57‐46.33	<.001
Histological grade
High (reference)	1	‐	‐
Low/intermediate	3.29	1.08‐12.10	.045
Histological subtype
Non‐mucoepidermoid carcinoma (reference)	1	‐	‐
Mucoepidermoid carcinoma	4.80	1.20‐19.15	.026

Abbreviation: CNB, core‐needle biopsy; FNAC, fine‐needle aspiration cytology.

Female sex (OR: 3.52; 95% CI: 0.89‐13.97; P = .074) and no smoking history (OR: 3.94; 95% CI: 0.95‐16.33; P = .059) were suggested as potential risk factors for preoperative undiagnosis, but statistical significance was not reached. Other characteristics, such as age, alcohol history, Charlson comorbidity index, and primary site, were not associated with an increased risk of preoperative undiagnosis of MSGC.

Treatment Strategies and Pathological Stage

Treatment strategies and pathological stages are summarized in Table 4. The median interval from preoperative diagnosis to surgery was 60 days and 24 days in the undiagnosed and diagnosed groups, respectively (P < .001).

Table 4.

Treatment Strategies and Pathological Stage

	Diagnosed (n = 49)	Undiagnosed (n = 73)	P‐value
Elapsed time from preoperative diagnosis to surgery (days)	24 (IQR, 23; range, 8‐129)	60 (IQR, 58; range 8‐365)	<.001
Extent of surgery
Primary
Parotid
Subtotal/total parotidectomy	21 (42.9%)	17 (23.3%)	.001
Partial parotidectomy	13 (26.5%)	45 (61.6%)
Submandibular gland
Gland excision	15 (30.6%)	11 (15.1%)	‐
Neck
None	13 (26.5%)	57 (78.1%)	<.001
Therapeutic neck dissection	17 (34.7%)	1 (1.4%)
Elective neck dissection	19 (38.8%)	15 (20.5%)
Pathological stage
T category
1	11 (22.5%)	24 (32.9%)	.522
2	12 (24.5%)	16 (21.9%)
3	13 (26.5%)	20 (27.4%)
4	13 (26.5%)	13 (17.8%)
N category
0	31 (63.3%)	69 (94.5%)	<.001
1	3 (6.1%)	3 (4.1%)
2	9 (18.4%)	1 (1.4%)
3	6 (12.2%)	0 (0.0%)
Overall stage
1	8 (16.3%)	24 (32.9%)	.028
2	10 (20.4%)	16 (21.9%)
3	11 (22.5%)	20 (27.4%)
4	20 (40.8%)	13 (17.8%)
Advanced stage	31 (63.3%)	33 (45.2%)	.050
Adverse pathololgical findings
Positive resection margin	19 (38.8%)	20 (27.4%)	.186
Lymphovascular invasion	17 (34.7%)	2 (2.7%)	<.001
Perineural invasion	18 (36.7%)	13 (17.8%)	.019
Extranodal extension	10 (20.4%)	1 (1.4%)	<.001
Postoperative adjuvant treatment
Yes	35 (71.4%)	41 (56.2%)	.088
No	14 (28.6%)	32 (43.8%)

Abbreviation: IQR, interquartile range.

For patients with MSGC of the parotid gland, near‐total/total parotidectomy was performed in 17 (23.3%) and 21 (42.9%) patients in the undiagnosed and diagnosed groups, respectively, while partial parotidectomy was conducted in 45 (61.6%) and 13 (26.5%) patients in the undiagnosed and diagnosed groups, respectively (P < .001).

With respect to neck dissection, no neck dissection, therapeutic neck dissection, and elective neck dissection were performed in 57 (78.1%), 1 (1.4%), and 15 (20.5%) patients in the undiagnosed group, respectively, compared to 13 (26.5%), 17 (34.7%), and 19 (38.8%) patients in the diagnosed group, respectively (P < .001).

With respect to pathological stage after surgery, 33 patients (45.2%) in the undiagnosed group were categorized as advanced stage, compared to 31 patients (63.3%) in the diagnosed group (P = .050). In terms of adverse pathological findings, lymphovascular invasion (2.7% vs 34.7%, P < .001), perineural invasion (17.8% vs 36.7%, P = .019), and extranodal extension (1.4% vs 20.4%, P < .001) were significantly less frequent in the undiagnosed group than in the diagnosed group, whereas positive resection margins were comparable between the 2 groups.

Postoperative adjuvant treatment, including radiation therapy or chemoradiation therapy, was administered to 41 (56.2%) and 35 (71.4%) patients in the undiagnosed and diagnosed groups, respectively (P = .088).

LRC

The 5‐year LRC was evaluated as a function of the preoperative diagnosis alone, grade‐matched model, and stage‐matched model.

The 5‐year LRCs were 91.3% and 86.2% in the diagnosed and undiagnosed groups, respectively (P = .495) (Figure 1A). In the grade‐matched model, the 5‐year LRCs were 100%, 94.0%, 89.6%, and 77.5% in the low/intermediate‐grade‐diagnosed, low/intermediate‐grade‐undiagnosed, high‐grade‐diagnosed, and high‐grade‐undiagnosed groups, respectively (P = .080) (Figure 1B). For the stage‐matched model, the 5‐year LRCs were 87.5%, 97.4%, 85.2%, and 71.8% in the early‐stage‐diagnosed, early‐stage‐undiagnosed, advanced‐stage‐diagnosed, and advanced‐stage‐undiagnosed groups, respectively (P = .003) (Figure 1C).

Figure 1.

Kaplan‐Meier curves for overall (A), grade‐matched (B), and stage‐matched (C) loco‐regional control according to the preoperatively diagnosed and undiagnosed groups.

Risk Factors for Treatment Failure in MGSC

Table 5 summarizes the results of the multivariate Cox regression analysis of possible risk factors for treatment failure in MSGC. Charlson comorbidity index ≥3 (HR: 7.13; 95% CI: 1.75‐29.01; P = .006), high‐grade histology (HR: 8.80; 95% CI: 1.70‐45.60; P = .010), positive resection margins (HR: 5.73; 95% CI: 1.45‐22.64; P = .013), perineural invasion (HR: 5.23; 95% CI: 1.63‐16.80; P = .006), and no adjuvant treatment (HR: 4.77; 95% CI: 1.64‐13.86; P = .004) were identified as independent risk factors for treatment failure.

Table 5.

Multivariate Analysis of Risk Factors for Treatment Failure in Major Salivary Gland Cancer

Covariates	Hazard ratio	95% confidence interval	P‐value
Sex
Female (reference)	1	‐	‐
Male	1.25	0.41‐3.83	.700
Age
<65 years (reference)	1	‐	‐
≥65 years	2.70	0.86‐8.51	.089
Smoking history
No (reference)	1	‐	‐
Yes	1.23	0.33‐4.63	.755
Charlson comorbidity index
1–2 (reference)	1	‐	‐
≥3	7.13	1.75‐29.00	.006
Preoperative undiagnostic status	2.58	0.67‐9.87	.167
Diagnosis (reference)	1	‐	‐
Undiagnosis	2.58	0.67‐9.87	.167
Histological grade
Low/intermediate (reference)	1	‐	‐
High	8.80	1.70‐45.60	.010
Disease stage
Early (reference)	1	‐	‐
Advanced	2.49	0.49‐12.60	.270
Positive resection margin
No (reference)	1	‐	‐
Yes	5.73	1.45‐22.64	.013
Lymphovascular invasion
No (reference)	1	‐	‐
Yes	2.15	0.49‐9.39	.310
Perineural invasion
No (reference)	1	‐	‐
Yes	5.23	1.63‐16.80	.006
No adjuvant therapy
No (reference)	1	‐	‐
Yes	4.78	1.64‐13.86	.004

Age ≥65 years (HR: 2.70; 95% CI: 0.86‐8.51; P = .089) and preoperative undiagnosis of MSGC (HR: 2.58; 95% CI: 0.67‐9.87; P = .167) were suggested as potential risk factors for treatment failure, but statistical significance was not reached.

Discussion

This study demonstrated that preoperatively undiagnosed MSGC was associated with distinct clinicopathological characteristics, treatment strategies, and clinical outcomes compared to preoperatively diagnosed MSGC.

Among the 73 patients in the undiagnosed group, 47.9% of MSGCs were misdiagnosed as benign disease on preoperative FNAC or CNB. This finding may be attributed to the higher proportion of low/intermediate‐grade disease in the undiagnosed group (75%) than in the diagnosed group (26.5%). Low/intermediate‐grade MSGCs often exhibit relatively bland pathological features, making preoperative diagnosis more challenging than in high‐grade MSGCs. ¹⁴ , ¹⁸ , ¹⁹ A previous study of 112 patients with low/intermediate‐grade parotid cancer reported benign or nondiagnostic FNAC results in 44.6% of cases, significantly higher than in high‐grade cancers (15.3%). ¹⁸ Another study involving 48 patients with salivary gland cancer showed that a preoperative diagnosis of malignancy was suggested in only 0% to 14.3% of low‐grade tumors, including adenocarcinoma, adenoid cystic carcinoma, and MEC, compared with 66.7% to 100% of high‐grade tumors, such as SDC and squamous cell carcinoma. ¹⁴ Consistent with these reports, MEC was the most frequent histologic subtype in the undiagnosed group, whereas SDC predominated in the diagnosed group. In addition, the diagnosed group showed a higher prevalence of aggressive pathological features, including lymphovascular invasion, perineural invasion, extranodal extension, and advanced disease stage. Taken together, these findings indicate that accurate preoperative diagnosis of MSGC is strongly dependent on its histological grade and subtype, underscoring the distinct pathological profile of preoperatively undiagnosed MSGCs.

Multivariate logistic regression analysis confirmed the impact of histological grade and subtype on preoperative diagnosis, demonstrating that low/intermediate‐grade histology and MEC were associated with a 3.3‐ and 4.8‐fold higher risk of preoperative undiagnosis of MSGC, respectively. A higher likelihood of preoperative undiagnosis was also observed among non‐smokers, which may be related to the previously reported association between nonsmoking status and low‐grade histologic subtypes, particularly MEC. ⁸ , ²⁰ In addition, the use of FNAC as the preoperative diagnostic modality was associated with a 12.9‐fold higher risk of preoperative undiagnosis compared with CNB. These results suggest that the choice of diagnostic modality is as critical as intrinsic tumor characteristics for achieving an accurate preoperative diagnosis of MSGC. The diagnostic superiority of CNB over FNAC has been consistently demonstrated, particularly with respect to lower nondiagnostic rates and higher sensitivity. ¹¹ , ¹² , ¹³ , ²¹ , ²² Our findings further support this evidence by demonstrating the increased risk of preoperative undiagnosis associated with FNAC.

Regarding treatment strategy, the undiagnosed group experienced a significantly longer interval from diagnosis to surgery than the diagnosed group (median, 60 vs 24 days). Although no studies have specifically evaluated the impact of the diagnosis‐to‐treatment interval (DTI) in MSGC, prior studies in HNC commonly define delayed DTI as treatment initiation beyond 30 days after diagnosis and have demonstrated an association between prolonged DTI and worse survival outcomes. ²³ , ²⁴ , ²⁵ Accordingly, timely initiation of treatment within 30 days is generally recommended in clinical practice for HNC. ²³ , ²⁴ , ²⁵ With respect to surgical management, less extensive surgery was more frequently performed in the undiagnosed group. Although limited resection without elective neck dissection may be appropriate for selected patients with low‐grade, early‐stage MSGC, this approach is insufficient for high‐grade or advanced‐stage disease, for which current guidelines recommend more extensive surgical treatment. ¹ , ⁸ , ¹⁷ , ²⁶ Taken together, the combination of delayed surgery and less extensive initial treatment suggests that preoperatively undiagnosed MSGC may be associated with less favorable oncologic outcomes than diagnosed cases, when histological grade and disease stage are comparable.

However, as noted above, the undiagnosed group was more frequently characterized by low/intermediate‐grade tumors, earlier disease stage, and fewer adverse pathological features, which complicates the assessment of the independent impact of accurate preoperative diagnosis on treatment outcomes. In fact, the 5‐year overall LRC rate did not differ between the diagnosed and undiagnosed groups in the present study. This observation is in line with 2 previous studies evaluating oncologic outcomes in patients with preoperatively undiagnosed salivary gland cancer, which likewise reported comparable LRC despite differences in preoperative diagnostic status. ² , ¹⁶ In 1 study of 44 MSGCs preoperatively misdiagnosed as benign, disease‐free survival was comparable to that of 44 MSGCs preoperatively diagnosed as malignant. ² Another study of 85 patients with preoperatively unexpected parotid cancer reported excellent oncologic outcomes, with 5‐year overall survival and relapse‐free survival rates of 100.0% and 95.2%, respectively. ¹⁶ However, neither study adjusted for histological grade or disease stage when evaluating oncologic outcomes according to preoperative diagnostic status. In contrast, the current study evaluated grade‐ and stage‐matched LRC to account for these potential confounders and identified a trend toward worse LRC in the undiagnosed group among patients with high‐grade or advanced‐stage MSGC. These results suggest that the clinical impact of accurate preoperative diagnosis may be more pronounced in high‐grade or advanced‐stage MSGCs compared to low/intermediate‐grade and early‐stage MSGCs.

Traditionally, tumor‐ and treatment‐related factors, including histological grade, disease stage, surgical margin status, perineural invasion, lymphovascular invasion, and adjuvant therapy, have been well established as prognostic factors, whereas patient‐related factors, such as age, sex, and comorbidity burden, have also been suggested to influence oncologic outcomes. ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² Our findings are consistent with this framework, demonstrating that a Charlson comorbidity index ≥3, high‐grade histology, positive resection margins, perineural invasion, and the absence of adjuvant therapy were independently associated with an increased risk of treatment failure. Although preoperative undiagnosis of MSGC—the primary variable of interest—was associated with an approximately 2.6‐fold higher risk of treatment failure, this association did not reach statistical significance in multivariate analysis. This finding may be attributable to the limited sample size and the inherently less aggressive tumor characteristics of preoperatively undiagnosed MSGCs, which likely introduced residual confounding despite adjustment for established prognostic factors.

This study has several methodological limitations. First, the selection of FNAC versus CNB was not randomized but based on clinical judgment, introducing the potential for selection bias in diagnostic sampling. In addition, preoperative diagnostic status was defined solely by FNAC or CNB results to ensure consistent cohort classification, which may not fully reflect real‐world multidisciplinary clinical practice. Second, although the study included 122 patients with MSGC, a relatively large cohort for this rare malignancy, subgroup analyses were limited by small sample sizes, reducing statistical power for LRC analysis and multivariate Cox regression modeling. Third, as this was a single‐center study, the generalizability of the findings may be limited.

Despite these limitations, the present study provides a comprehensive evaluation of the clinical significance of preoperative diagnosis in MSGC, particularly with respect to clinicopathological characteristics, treatment strategies, and oncologic outcomes. Furthermore, the grade‐ and stage‐matched outcome analysis offers important insights into the oncologic impact of preoperative undiagnosis, especially in high‐grade or advanced‐stage MSGCs.

Conclusion

This study highlights the differences in clinicopathological characteristics, treatment strategies, and oncological outcomes between preoperatively diagnosed and undiagnosed MSGCs. Preoperative undiagnosis was associated with low/intermediate‐grade tumors, reliance on FNAC, and less extensive surgical interventions. Grade‐ and stage‐matched LRC revealed a trend towards worse oncological outcomes in undiagnosed high‐grade or advanced‐stage MSGCs. These findings highlight the clinical importance of accurate preoperative diagnosis in optimizing the overall management process for MSGC.

Author Contributions

Dongbin Ahn, design, patient enrolment, data analyses, manuscript drafting, editing, and final approval; Joseph Califano, editing, critical revision, and final approval; Jihye Kwak, patient enrolment, data analyses, and critical revision; Sung Jae Heo, patient enrolment, data analyses, and critical revision; Kyu‐Yup Lee, design, patient enrolment, interpretation, editing, critical revision, and final approval.

Disclosures

Competing interests

None.

Funding source

None.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.