Tailored Mandibular Advancement Therapy Guided Through a Mandibular Positioner: Predictive Value in Obstructive Sleep Apnea

Abstract

Objective

Mandibular advancement devices (MADs) are a recognized alternative to continuous positive airway pressure (CPAP) for treating obstructive sleep apnea (OSA), though patient selection is still challenging. Tailored MAD guided through drug induced sleep endoscopy (DISE) with a mandibular titratable positioner (SAM) offers a stable and reproducible simulation of MAD effects during DISE, potentially enhancing outcomes after this DISE‐SAM protocol. We aim to assess its positive predictive value (PPV) in patients treated with MAD by analyzing changes in AHI, minimum oxygen saturation, and CT90, considering OSA severity and recommendations based on the expected response.

Study Design

Prospective cohort study.

Setting

Academic secondary medical centers.

Methods

A prospective, multicenter study was conducted including 77 OSA patients across 3 hospitals. After DISE with the SAM device, 66 patients concluded MAD therapy based on SAM recommendations: Single use (S), Alternative (A), or Multiple treatment approach (M). Pretreatment and posttreatment sleep studies measured AHI, Min O₂ Sat, and CT90. Response was analyzed with standard AHI‐oxygen saturation‐based criteria and newly tailored criteria per SAM recommendation category.

Results

MAD therapy led to significant improvement in AHI (from 32.26 to 7.19), min O₂ Sat (from 80.83% to 85.87%), and CT90 (from 11.33% to 3.99%). Stratified analyses showed significant treatment effects in moderate and severe OSA groups and across SAM recommendations. The MAD therapy guided through DISE‐SAM protocol demonstrated high PPV, also when using tailored success criteria.

Conclusions

The DISE‐SAM protocol demonstrated a high positive predictive value and significant improvements in key respiratory parameters among the selected patients using MAD.

Obstructive sleep apnea (OSA) is a prevalent medical condition that significantly impacts both health and quality of life. ¹ , ² Early diagnosis and accurate upper airway topodiagnosis are essential for selecting appropriate candidates for targeted interventions, ³ optimizing outcomes, and improving long‐term treatment efficacy while reducing the risk of inadequate response and persistent untreated OSA.

Although continuous positive airway pressure (CPAP) is considered the gold standard treatment for OSA patients, mandibular advancement devices (MAD) are becoming an increasingly popular alternative worldwide; however, treatment response remains highly variable among patients. Improving patient selection and, consequently, treatment efficacy, could support the broader acceptance of MAD therapy as a first‐line treatment option. ⁴ Among all methods for the prediction of MAD treatment outcome, DISE provides a dynamic evaluation of the upper airway during sedation, ⁵ enabling real‐time assessment of specific obstruction sites. This could be crucial for tailoring treatment, ⁶ as assessing upper airway modifications while simulating the effect of a MAD may facilitate patient selection for this therapy. ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴

Tailored MAD guided through DISE with a mandibular titratable positioner, the Selector de Avance Mandibular (SAM) offers a stable and reproducible simulation of MAD effects during DISE, as proposed in the DISE‐SAM protocol. ¹³ It allows the simulation of the effect of a MAD and assessment of its impact on upper airway collapsibility.

This approach supports the personalization of MAD therapy by allowing clinicians to tailor treatment based on predicted responses, as indicated by the SAM recommendations: single use (S), when a complete response is expected; alternative use (A), when only a partial response is anticipated; or multiple use (M), when a combination of treatments is required, regardless of OSA severity. ¹³

Our primary objective was to evaluate the clinical success of MAD treatment in patients selected using the DISE‐SAM protocol, by analyzing changes in apnea‐hypopnea index (AHI), minimum oxygen saturation (Min O₂ Sat), and percentage of total recording time during which blood oxygen saturation remains below 90% (CT90). These outcomes were studied concerning both OSA severity and SAM‐based treatment recommendations. In addition, the study aimed to calculate the PPV of the DISE‐SAM protocol to guide MAD treatment decisions.

Material and Methods

Study Design

This multicenter prospective observational study included a baseline clinical evaluation and sleep study, followed by DISE performed with the DISE‐SAM protocol to assign each participant to a SAM recommendation category. Eligible patients then initiated MAD treatment. After a period of adaptation and titration, a follow‐up sleep study was conducted to assess post‐treatment respiratory outcomes. Data analysis comprised, firstly, the outcome assessment for the whole cohort, subgroup analyses by OSA severity and SAM category and, secondly, evaluation of the PPV of SAM recommendations using established and newly created success criteria (Figure 1).

Figure 1.

Overview of the study design, including baseline assessment, DISE‐SAM evaluation and SAM category assignment, MAD treatment, follow‐up sleep study, and subsequent outcome and predictive value analyses based on oxygenation improvement and AHI reduction. AHI, apnea‐hypopnea index; DISE, Drug Sleep endoscopy; MAD, mandibular advancement devices; PPV, positive predictive value; SAM, Selector de Avance Mandibular.

Participants

OSA patients were recruited from 3 hospitals (Fuenlabrada, Fundación Jiménez Diaz, La Zarzuela) in Madrid, Spain. Patients with mild to severe OSA were consecutively included if they were over ¹⁵ 18 years old, had undergone evaluation with the DISE‐SAM protocol, and had received a SAM recommendation for MAD therapy. Exclusion criteria comprised prior OSA surgery, dental or periodontal limitations, severe temporomandibular disorders, limited mandibular advancement, inability to use MADs, craniofacial anomalies, muscle relaxant use, and REM‐related OSA (AHI during REM sleep at least twice the AHI during non‐REM sleep), as propofol‐induced DISE does not reproduce REM sleep and may compromise the predictive accuracy of the SAM assessment.

Of the 77 patients enrolled, 66 completed the study (4 discontinued MAD use and 7 were lost to follow‐up). Among them, 10 had mild (AHI: 5‐14.9), 27 moderate (AHI: 15‐29.9), and 29 severe (AHI ≥30) OSA. Based on the DISE‐SAM protocol, 15 patients received an S recommendation, 34 an A recommendation and 17 an M recommendation. Baseline characteristics are described in Table 1.

Table 1.

Baseline Characteristics of the Overall Cohort and of Subgroups Defined by Severity and SAM Recommendation

	n (m; f)	Median age (CI)	Median BMI (CI)
Whole group	66 (54; 12)	54 (52‐55)	25.8 (25.1‐26.3)
Severity
Mild	10 (8; 2)	56 (54.5‐66)	25.55 (24.3‐26.3)
Moderate	27 (21; 6)	54.5 (51‐58)	26.3 (24.21‐26.8)
Severe	29 (25; 4)	53 (52‐54)	25.7 (25‐27)
SAM recommendation
S	15 (11; 4)	63 (56‐63)	24.9 (23‐26.4)
A	34 (29; 5)	53 (51.16‐55)	26.2 (25.7‐27)
M	17 (14; 3)	55.5 (52‐57)	25.59 (24‐27.8)

Abbreviations: BMI, body mass index; CI, confidence interval; f, females; m, males.

Procedure

The study was approved by the Hospital Universitario de Fuenlabrada Ethics Committee (APR 20/30) and was conducted in accordance with the principles of the Declaration of Helsinki. ¹⁵ All participants provided informed consent and data were collected anonymously.

All patients underwent either standard polysomnography or level 2 sleep study. All centers adhered to standardized AASM criteria and calibration procedures performed by trained technologists at each site. Key respiratory parameters were recorded, including the AHI, Min O₂ sat, and CT90.

DISE was performed following the European position paper  ⁶ using target‐controlled infusion of 2% propofol (BIS 50‐70) and continuous monitoring of oxygenation and cardiorespiratory parameters while avoiding vasoconstrictors and topical anesthetics. A flexible nasendoscope was inserted transnasally to directly visualize dynamic upper‐airway behavior during simulated sleep. Collapse patterns and severity were systematically assessed at multiple levels (velopharynx, oropharynx, tongue base and epiglottis) using the standardized VOTE classification. ¹⁶ The SAM device with the trays adapted to the dental arches (Figure 2A and B) was employed to simulate the MAD effect during DISE. According to the observed modifications in the upper airway, patients were classified into 3 groups. S: complete collapses were eliminated at all levels, a complete response is expected, leading to a single use recommendation. A: an improvement in collapsibility ≥50% with no residual complete collapses suggests a partial response. In this group, MAD therapy is recommended as a sole treatment and is considered preferable to no treatment leading to an alternative recommendation. M: partial improvement with residual complete collapses indicates the need for multiple treatment: MAD + positional therapy (n = 10), MAD + pharyngoplasty (n = 5), MAD + positional therapy + myofunctional therapy (n = 2). All investigators involved in the study had experience in DISE. In addition, the same dentist was present at all procedures and performed the maneuvers.

Figure 2.

(A) Trays inserted in the SAM device; (B) SAM device simulating the effect of MAD during DISE; (C) SAM titration; (D) SAM device with forks; (E) minimum jaw opening; (F) increased jaw opening. DISE, Drug Sleep endoscopy; MAD, mandibular advancement devices; SAM, Selector de Avance Mandibular.

Mandibular position records were obtained in all patients for MAD fabrication using the SAM device, which employs forks instead of trays to allow natural joint and muscle motion and multidirectional mandibular movement (Figure 2D and F). Treatment was initiated at 60% of maximum protrusion to facilitate adaptation, while jaw opening was kept to the minimum necessary to reduce appliance bulk, improve lip seal and comfort, ¹⁷ and facilitate protrusive movement, ¹⁸ increasing vertical opening only when required by dental or occlusal morphology (Figure 2F). The MADs utilized in this study included the SOMNOMED FLEX (32 patients), SOMNOMED AVANT (29 patients), and NOA‐Orthoapnea (5 patients) designs and were chosen according to dentition, occlusion, history of temporomandibular disorders or bruxism.

After titration and 1 month of continuous MAD use, control sleep studies were performed to record AHI, Min O₂ sat, and CT90 with the MAD in situ.

Treatment Response Definitions

According to Sutherland et al. ¹⁹ 3 definitions of treatment success based exclusively on the AHI have been proposed: posttreatment AHI < 5/h, indicating complete resolution of OSA (definition 1); posttreatment AHI < 10/h combined with a ≥50% reduction from baseline AHI, representing complete to very mild residual OSA (definition 2); and a ≥50% reduction in AHI from baseline (definition 3). However, several authors have argued that treatment success should not be defined solely by changes in AHI. ²⁰ , ²¹ Pang et al. ²⁰ proposed the ‘SLEEP GOAL' concept, which defines treatment success to surgical treatment based on multiple parameters, including an improvement of at least 50% in CT90 and a reduction in AHI greater than 50%. In line with these recommendations and considering that the primary goal of the present study is to assess the PPV of the DISE‐SAM protocol in reducing respiratory events, the newly added success criteria based on expected physiological outcomes derived from the anatomical and functional changes observed during DISE included not only improvements in AHI but also in oxygen saturation. Given the differing expectations by SAM category, success criteria were adapted accordingly. For “S” recommendation a stringent success criteria was applied, requiring ≥50% AHI reduction plus posttreatment AHI < 10 and CT90 < 10%, For “A” recommendation, improvement was defined as ≥50% reduction in both AHI and CT90, reflecting the expectation of a partial response. In the ‘M' group, where outcomes rely on combined therapies, success could include achieving ≥50% AHI reduction, plus AHI < 10 and CT90 < 10% if adjunctive treatments also succeeded. This approach aims to facilitate the interpretation of DISE findings and support clinical decision‐making regarding whether MAD therapy should be considered as a first‐line treatment, a rescue option, or part of a combined therapeutic strategy.

Statistical Analysis

Data normality and statistical assumptions were checked with Shapiro‐Wilk's test. Considering the nonnormal distribution of several variables, Wilcoxon signed‐rank tests were used to analyze changes in parameters due to MAD treatment.

First, the entire cohort was analyzed by comparing pretreatment and posttreatment AHI, min O₂ sat, and CT90. Next, participants were stratified into 3 groups according to severity of the AHI—mild (AHI: 5‐14.9), moderate (AHI: 15‐29.9), and severe (AHI ≥ 30)—and pretreatment and posttreatment changes were analyzed for the same measures. Subsequently, participants were grouped according to their DISE‐SAM recommendations, and again pretreatment and posttreatment changes were evaluated.

Sex distribution across the severity and the SAM recommendation groups was compared using χ ² tests. Differences in age and BMI among these groups were analyzed using the Kruskal‐Wallis test. The relationship between SAM recommendation and severity was also examined by χ ² analysis. The potential presence of differences according to the type of MAD device used was analyzed using the Kruskal‐Wallis test.

Additionally, the PPV of the SAM recommendation was also calculated according to the definitions by Sutherland et al., ¹⁹ Pang et al., ²⁰ and the newly introduced specific success criteria for S, A, and M groups.

Given the nonnormality of some variables and the relatively small size of some groups, median values and their 95% confidence intervals were estimated using a bootstrap procedure with 1000 resamples (bias‐corrected and accelerated method) to yield robust estimates.

A P < .05 was considered significant for all tests. Analyses were performed using SPSS v27.0, while effect sizes and power calculations were conducted using G*Power. ²² , ²³

Results

First, a significant improvement was observed in the 3 key measures (AHI, min O₂ sat, and CT90) when comparing pre‐and post‐treatment data for the entire cohort (Table 2 and Figure 3).

Table 2.

Baseline Versus MAD Treatment Results for AHI, Min O₂ Sat, and CT90 for the Overall Cohort

	n	Premedian (CI)	Postmedian (CI)	Wilcoxon Z	P	Effect size	Power (1−β)
AHI	66	27.95 (24.8‐33.8)	5.1 (4‐7.4)	−7.04	<.001	1.65	1
Min O2 Sat	66	81.5 (80‐84)	87.5 (86‐88)	−5.44	<.001	0.86	1
CT90	66	6.75 (3.3‐9.8)	0.6 (0.35‐1)	−5.29	<.001	0.56	0.99

Abbreviations: AHI, apnea‐hypopnea index; CI, confidence interval; MAD, mandibular advancement devices; Min O₂ Sat, minimum oxygen saturation.

Figure 3.

Pretreatment and posttreatment AHI, min O₂ sat, and CT90 with MAD therapy for the overall cohort. Median values and CIs are represented. AHI, apnea‐hypopnea index; CI, confidence interval; MAD, mandibular advancement devices; Min O₂, minimum oxygen saturation.

Then, participants were stratified into 3 groups based on AHI severity. Descriptive statistics and results for each group are shown in Table 3 and Figure 4. Significant differences were observed across all measures in the moderate and severe groups (p < 0.001 in all cases). In the mild OSA group, a significant difference was found for AHI (p = 0.01), while the difference for CT90 was marginally significant (p = 0.07), and non‐significant for min O₂ Sat (p = 0.22).

Table 3.

Comparison of Baseline and Post‐MAD Treatment AHI, Min O₂ Sat, and CT90 in Mild, Moderate, and Severe groups

	Measure	Premedian (CI)	Postmedian (CI)	Wilcoxon Z	P	Effect size	Power (1−β)
Mild (n = 10)	AHI	12.35 (9.9‐14.1)	4 (4‐4)	−2.5	.01	1.48	0.98
	Min O2 Sat	85 (85‐85)	86.5 (82‐89)	−1.23	.22	0.31	0.14
CT90	3 (0.99‐25.6)	0.85 (0‐4.6)	−1.79	.07	0.18	0.08
Moderate (n = 27)	AHI	24.3 (20.55‐27)	3.99 (3‐5.6)	−4.54	<.001	4.29	1
	Min O2 Sat	82 (80‐84)	89 (89‐89)	−4.03	<.001	0.95	1
CT90	3 (1.15‐8.82)	0.1 (0‐1.32)	−3.2	.001	0.57	0.79
Severe (n = 29)	AHI	42.7 (40.3‐52.7)	8.5 (5.8‐12.9)	−4.7	<.001	3.14	1
	Min O2 Sat	80 (79‐80)	85.5 (83‐87)	−3.46	<.001	0.94	1
	CT90	10 (7.9‐10.55)	1 (0.4‐1.5)	−3.87	<.001	0.78	0.98

Abbreviations: AHI, apnea‐hypopnea index; CI, confidence interval; MAD, mandibular advancement devices; Min O₂ Sat, minimum oxygen saturation.

Figure 4.

Baseline and posttreatment AHI, Min O₂ Sat, and CT90 after MAD therapy for the mild, moderate, and severe groups. Median values and CIs are represented. AHI, apnea‐hypopnea index; MAD, mandibular advancement devices; Min O₂ Sat, minimum oxygen saturation.

Participants were also categorized into 3 groups according to SAM recommendations. Descriptive statistics and results for each group are shown in Table 4 and Figure 5. Significant differences were observed across all measures in all 3 groups, except for CT90 in the S recommendation group (P = .099).

Table 4.

Comparison of Baseline and Post‐MAD Treatment AHI, Min O₂ Sat, and CT90 in S, A, and M Groups

	Measure	Premedian (CI)	Postmedian (CI)	Wilcoxon Z	P	Effect size	Power (1−β)
S (n = 15)	AHI	27 (19.6‐34.7)	3.8 (3.2‐5.8)	−3.35	<.001	1.32	1
	Min O2 Sat	83 (80‐85)	87 (82‐89)	−2.07	.039	0.59	0.55
CT90	7.1 (3.9‐12)	1.1 (0.4‐2.9)	−1.65	.099	0.18	0.1
A (n = 34)	AHI	29.2 (25.3‐40.3)	5.9 (4‐10.38)	−5.09	<.001	1.9	1
	Min O2 Sat	83 (80‐84)	87 (85.5‐88)	−3.72	<.001	0.8	0.99
CT90	7.15 (1.5‐10)	0.69 (0.35‐2.1)	−3.71	<.001	0.52	0.83
M (n = 17)	AHI	27.7 (19‐30.6)	6.6 (4‐9.35)	−3.57	<.001	1.53	1
	Min O2 Sat	79 (78‐85)	89 (88‐89)	−3.34	<.001	1.22	1
	CT90	4 (1.25‐16)	0 (0‐0.7)	−3.62	<.001	0.73	0.79

Abbreviations: AHI, apnea‐hypopnea index; CI, confidence interval; MAD, mandibular advancement devices; Min O₂ Sat, minimum oxygen saturation.

Figure 5.

Pretreatment and posttreatment AHI, Min O₂ Sat, and CT90 after MAD therapy for the S, A and M groups. Median values and CIs are represented. AHI, apnea‐hypopnea index; CI, confidence interval; MAD, mandibular advancement devices; Min O₂, minimum oxygen saturation.

The PPV of the SAM recommendations for treatment success was subsequently calculated (see Table 5).

Table 5.

Comparison of the PPV of the SAM Recommendations Depending on the Success Criteria Applied

	Definition 1 AHI < 5/h	Definition 2 AHI < 10/h plus ≥ 50% reduction	Definition 3 ≥ 50% AHI reduction	SLEEP GOAL AHI reduction by 50% plus CT 90 improvement by 50%	S Criteria AHI < 10/h plus ≥ 50% reduction plus CT90 < 10	A Criteria AHI reduction by 50% plus CT 90 improvement by 50%	M Criteria AHI < 10/h plus ≥ 50% reduction plus CT90 < 10
S	73.33%	86.67%	93.33%	84.62%	86.67%
A	47.06%	61.76%	85.29%	84.85%		84.85%
M	35.29%	70.59%	94.12%	94.12%			70.59%

Definition 1, 2 & 3: Sutherland et al. ²⁰ success criteria based on AHI. SLEEP GOAL: CT 90 (O) and AHI (A) based on oxygenation improvement and AHI reduction success criteria from Pang et al. ¹⁹ Success criteria for S, A, and M recommendations.

Abbreviations: A, alternative treatment recommendation; M, multiple treatment approach recommendation; S, single‐use recommendation.

Finally, considering the distribution of the baseline parameters and to rule out the potential confounding effects of age, sex, OSA severity, and SAM recommendations, several analyses were conducted. Across the 3 severity levels, no significant differences were observed in age (H = 4.64, P = .10), BMI (H = 0.43, P = .81), or sex distribution (χ ²= 0.69, P = .71). Likewise, no association was found between OSA severity and SAM recommendations (χ ²= 1.32, P = .86). Regarding the groups formed based on SAM recommendation, no significant differences were detected in age (H = 3.53, P = .17), BMI (H = 3.34, P = .19), or sex distribution (χ ²= 1.01, P = .61). In addition, Kruskal‐Wallis tests were conducted to examine whether there were significant differences in any parameter across groups treated with different MAD designs. No statistically significant differences were found (P > .05 in all cases).

Discussion

This study assessed the PPV of the DISE‐SAM protocol by comparing predicted and actual MAD outcomes. Specifically, it analyzed the PPV of the DISE‐SAM protocol in personalizing therapy and improving not only AHI, as explored in previous studies, ⁹ , ²⁴ but also Min O₂ Sat and CT90, both in the overall sample and stratified by severity and SAM recommendations. Patients showed improvements across these parameters, whether analyzed as a whole or by severity or by SAM‐based recommendations.

First, the treatment effect was examined in the overall cohort. Significant gains were found in AHI, Min O₂ Sat, and CT90 for the entire cohort. Second, when stratified by severity, robust effects were confirmed in moderate and severe OSA, supporting DISE's utility for selecting candidates for MAD treatment. ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ²⁴ , ²⁵ , ²⁶ These results reinforce that moderate/severe OSA patients derive the greatest benefit from this selection method. In mild OSA, large AHI effect sizes suggest a strong treatment effect and a meaningful physiological benefit. The absence of significant oxygenation improvements in the mild group is likely driven by smaller baseline impairment, and small effect sizes, although insufficient statistical power due to the small sample size cannot be excluded. When stratified by SAM recommendations, patients demonstrated consistent improvements across nearly all respiratory parameters, with predominantly high to very high effect sizes, supporting a strong and robust treatment effect with the DISE‐SAM guided approach. Although CT90 did not reach statistical significance in the S recommendation group, this is likely attributable to its smaller effect size, leading to a limited statistical power, rather than a lack of treatment effect. In contrast, the large effect size for AHI across all recommendation groups highlights its sensitivity and clinical relevance, reinforcing the strong treatment effect and the predictive value of DISE‐SAM in personalizing mandibular advancement therapy.

When the PPV of SAM‐based recommendations was analyzed, the DISE‐SAM protocol showed a high ability to identify patients likely to benefit from MAD therapy and to guide the choice between standalone and combination treatments. Using the criteria proposed by Sutherland et al., ¹⁹ the stringent Definition 1 yielded the lowest PPV, classifying many improving patients as nonresponders. Definition 2, with more lenient thresholds, produced PPV values aligned with expectations across SAM categories, although some improvement remained undetected, particularly in recommendation A, where only partial response is anticipated. Uniform success criteria were applied to all patients for comparability with previous studies, while SAM‐specific criteria were used solely to assess concordance between predicted and observed treatment response.

Applying the more permissive Definition 3 resulted in high PPV ²⁷ across all SAM groups. However, consistent with Pang et al. ²⁰ and Mediano et al., ²¹ treatment success should not rely solely on AHI as inclusion of CT90 improves clinical relevance. The SLEEP GOAL criteria also yielded high PPV across recommendations but their uniform allowance of partial response is suboptimal for recommendation S, where standalone MAD should achieve full response, and for recommendation M, where complete resolution is expected with effective combination therapy.

In contrast, SAM‐specific success criteria better reflect OSA heterogeneity and avoid the limitations of uniform thresholds. Using these tailored criteria, PPV differed across SAM categories, remaining high for recommendations S and A and lowest for M. In the M group, results confirmed that MAD alone is insufficient in persistent complete collapse and that outcomes depend on combination therapies, although ~70% still achieved success. For recommendation A, less strict criteria appropriately identify patients in whom MAD may serve as an effective rescue therapy. Overall, aligning expected treatment response with SAM recommendations supports more precise and clinically meaningful therapeutic decision‐making.

Despite the high positive predictive value of the DISE‐SAM protocol, some patients did not fully conform to its predictions. Some patients failed to achieve treatment success despite favorable SAM recommendations. Others classified as A achieved near‐complete resolution (AHI < 10) despite baseline multilevel collapse, likely due to widespread improvement in airway collapsibility under SAM. This implies that the extent of VOTE score ¹⁶ reduction under SAM, rather than baseline severity, may better predict outcomes. Accordingly, patients with baseline multilevel complete collapse and partial residual obstruction under SAM may have a better prognosis than those with baseline single‐level complete collapse that remains partially unresolved with SAM. Additionally, patients with mild AHI but high CT90 did not improve oxygenation despite airway stabilization, highlighting the pitfalls of relying solely on AHI. Yet in moderate/severe OSA with elevated CT90, SAM‐induced upper airway stabilization normalized respiratory metrics, indicating that although low CT90 predicts MAD success, ²⁷ high CT90 in these patients does not preclude benefit.

Consistent with prior studies ²⁹ , ³⁰ MAD design did not significantly influence treatment outcomes among responders, as all designs similarly improved AHI, CT90, and Min O₂ Sat. These findings highlight patient selection and titration as more critical factors than appliance design, although body position, REM‐related events, and jaw opening should be considered during titration due to their potential impact on airway collapsibility.

Among the study's limitations, the use of propofol may not fully replicate natural sleep conditions, particularly in patients with predominantly REM‐related events. Positional dependency was intentionally not used to guide SAM recommendations, as positional OSA reflects sleep posture rather than the anatomical response of the upper airway to mandibular advancement, which is the target mechanism of MAD therapy. A larger sample of mild OSA patients would improve statistical power and oxygenation analysis. Nevertheless, consistent effect sizes across groups support the robustness and clinical relevance of the findings. As in most clinical studies, some variability must be acknowledged. While some patients predicted as responders during DISE‐SAM did not meet formal success criteria after treatment, others classified as non‐responders might still achieve favorable outcomes with MAD therapy. This discrepancy could be attributed to multiple factors, including the pharmacodynamic variability of the sedative agent and individual patient sensitivity, or subtle differences in anesthetic technique. These factors highlight the need for standardized protocols and careful sedation management to improve the predictive reliability of DISE‐based assessment. It is also essential to further investigate the negative predictive value of the DISE‐SAM protocol to better understand its ability to identify true non‐responders. Lastly, although all DISE procedures were performed by the same operator, a potential learning curve in the use of the SAM device cannot be excluded. As with any new technique, small improvements over time may have influenced consistency in maneuver execution and interpretation.

Conclusions

The DISE‐SAM protocol demonstrated a high PPV in identifying suitable candidates for MAD therapy, with significant improvements in AHI, Min O₂ Sat, and CT90. These benefits were observed across all OSA severity levels, with the strongest results in moderate and severe cases. The structured SAM recommendations allow an individualized strategy that improves patient selection, enhances treatment efficacy, and may reduce the need for additional interventions. This contributes to more effective and cost‐efficient care.

Author Contributions

Patricia Fernández‐Sanjuán, conception, design, data collection, data interpretation, manuscript writing; Gabriela Bosco, data collection, interpretation, manuscript writing; Nuria Pérez‐Martín, data collection, interpretation, manuscript writing; Marta Alcaraz, clinical data collection, interpretation, manuscript writing; Juan J. Arrieta, clinical data collection; Jaime Sanabria, clinical data collection; Silvia González‐Pondal, clinical data collection; Rodolfo Lugo, data interpretation, manuscript revision; Marcos Ríos‐Lago, data analysis, data interpretation, manuscript revision; Carlos O'Connor‐Reina, data interpretation, manuscript revision; Peter Baptista, data interpretation, manuscript revision; Guillermo Plaza, supervision, design, interpretation, manuscript writing, final approval.

Disclosures

This article was presented at the ISSS Annual Meeting, October 8th‐9th, 2025, Indianapolis, Indiana.

Competing interests

Patricia Fernández‐Sanjuán is the creator of the SAM device®.

Funding source

None.