ORIGINAL ARTICLE Pulse Wave Amplitude Drops Index: A Biomarker of Cardiovascular Risk in Obstructive Sleep Apnea Geoffroy Solelhac1*, Manuel Sa  nchez-de-la-Torre4,6,7*, Margaux Blanchard8,9*, Mathieu Berger1, Camila Hirotsu1, o Imler , Alicia Sa The 1  nchez-de-la-Torre4,6,7, Jose Haba-Rubio1, Nicola Andrea Marchi1, Virginie Bayon1, bastien Bailly , François Goupil11, Adrien Waeber1, Gre Se 10 gory Heiniger1, Thierry Pigeanne12, Esther Gracia-Lavedan , Andrea Zapater , Jorge Abad , Estrella Ordax14, Marıa Jose 5 4,6 13  Masdeu13, Valentin Cabriada-Nun ~o , Carlos Egea , Sandra Van Den Broecke , Peter Vollenweider2, 16 7,17 1,18 Pedro Marques-Vidal , Julien Vaucher2, Giulio Bernardi19, Monica Betta19, Francesca Siclari1,20,21, 2 Ferran Barbe7,5‡, Fre de ric Gagnadoux22,23‡, and Raphael Heinzer1,3‡ Abstract interval]), patients with a low PWAD index and OSA had a higher incidence of cardiovascular events compared with the Rationale: It is currently unclear which patients with obstructive high-PWAD and OSA group and those without OSA in the sleep apnea (OSA) are at increased cardiovascular risk. HypnoLaus cohort (HR, 2.16 [1.07–4.34], P = 0.031; and 2.35 [1.12–4.93], P = 0.024) and in the PLSC (1.36 [1.13–1.63], Objective: To investigate the value of pulse wave amplitude P = 0.001; and 1.44 [1.06–1.94], P = 0.019), respectively. In the drops (PWADs), reflecting sympathetic activations and ISAACC cohort, the low-PWAD and OSA untreated group had a vasoreactivity, as a biomarker of cardiovascular risk in OSA. higher cardiovascular event recurrence rate than that of the Methods: PWADs were derived from pulse oximetry–based no-OSA group (2.03 [1.08–3.81], P = 0.028). In the PLSC and photoplethysmography signals in three prospective cohorts: HypnoLaus cohorts, every increase of 10 events per hour in the HypnoLaus (N = 1,941), the Pays-de-la-Loire Sleep Cohort continuous PWAD index was negatively associated with incident (PLSC; N = 6,367), and “Impact of Sleep Apnea syndrome in the cardiovascular events exclusively in patients with OSA (HR, evolution of Acute Coronary syndrome. Effect of intervention with 0.85 [0.73–0.99], P = 0.031; and HR, 0.91 [0.86–0.96], P , 0.001, CPAP” (ISAACC) (N = 692). The PWAD index was the number of respectively). This association was not significant in the no-OSA PWADs (.30%) per hour during sleep. All participants were group and the ISAACC cohort. divided into subgroups according to the presence or absence of OSA Conclusions: In patients with OSA, a low PWAD index (defined as >15 or more events per hour or ,15/h, respectively, on reflecting poor autonomic and vascular reactivity was the apnea–hypopnea index) and the median PWAD index. Primary independently associated with a higher cardiovascular risk. outcome was the incidence of composite cardiovascular events. Measurements and Main Results: Using Cox models adjusted Keywords: sleep apnea; cardiovascular risk; PWAD; pulse for cardiovascular risk factors (hazard ratio; HR [95% confidence oximeter; pulse wave amplitude (Received in original form June 28, 2022; accepted in final form April 4, 2023) This article is open access and distributed under the terms of the Creative Commons Attribution Non-Commercial No Derivatives License 4.0. For commercial usage and reprints, please e-mail Diane Gern (dgern@thoracic.org). *Co–first authors. ‡ Co–last authors. The CoLaus study was supported by research grants from GlaxoSmithKline, the Faculty of Biology and Medicine of the University of Lausanne, and the Swiss National Science Foundation (grants 33CSCO-122661, 33CS30-139468, 33CS30-148401, and 32473B-182210). The HypnoLaus study received additional support from the “Ligue Pulmonaire Vaudoise,” the Lancardis Foundation, and the Foundation Leenaards. The ISAACC study was funded by the Institut de Salud Carlos III (PI10/02763, PI10/02745, PI18/00449, and PI19/00907), cofunded by FEDER “Una manera de hacer Europa,” SEPAR; the Catalonian Cardiology Society; ResMed, Ltd. (Australia); EsteveTeijin (Spain); Oxigen Salud (Spain); AllerGen; and CIBERES. M.S.-d.-l.-T. received financial support from a “Ramo n y Cajal” grant (RYC2019-027831-I) from the “Ministerio de Ciencia e Innovacio n – Agencia Estatal de Investigacio  n,” cofunded by the European Social Fund/“Investing in your future.” The Pays de la Loire Sleep Cohort is supported by research grants from the Pays de la Loire Respiratory Health Research Institute. F.S. is supported by the European Research Council Starting Grant 101039782 DREAMSCAPE. Am J Respir Crit Care Med Vol 207, Iss 12, pp 1620–1632, Jun 15, 2023 Copyright © 2023 by the American Thoracic Society Originally Published in Press as DOI: 10.1164/rccm.202206-1223OC on April 5, 2023 Internet address: www:atsjournals:org 1620 American Journal of Respiratory and Critical Care Medicine Volume 207 Number 12 | June 15 2023 ORIGINAL ARTICLE 1 Center for Investigation and Research in Sleep, 2Department of Medicine, Internal Medicine, and 3Pulmonary Department, Lausanne University Hospital and University of Lausanne, Lausanne, Switzerland; 4Precision Medicine in Chronic Diseases and 5Translation Research in Respiratory Medicine, University hospital Arnau de Vilanova-Santa Maria, IRB Lleida, Lleida, Spain; 6Department of Nurs- ing and Physiotherapy, Faculty of Nursing and Physiotherapy, University of Lleida, Lleida, Spain; 7Centro de Investigacio n Biome dica en Red de Enfermedades Respiratorias (CIBERES), Madrid, Spain; 8Ecole Supe rieur D’Electronique de l’Ouest, Angers, France; 9 Laboratoire d’Acoustique de l’Universite  du Mans, Le Mans, France; 10Universite  Grenoble Alpes, INSERM, CHU Grenoble Alpes, HP2, Grenoble, France; 11Department of Respiratory Diseases, Le Mans General Hospital, Le Mans, France; 12Unite  respiratoire, Po ^le sante des Olonnes, Olonne sur Mer, France; 13Respiratory Department, University hospital Germans Trias I Pujol, Barcelona, Spain; 14Respiratory Department, Burgos University hospital, Burgos, Spain; 15Respiratory and Sleep Department, University hospital  i Innovacio Parc Taulı, Institut Investigacio  Parc Taulı (I3PT), Universitat Autonoma de Barcelona, Sabadell, Spain; 16Respiratory Department, Cruces University hospital, Bizkaia, Spain; 17Servicio de Investigacio n OSI, Araba University hospital, IIS Bioaraba, Vitoria, Spain; 18Service de Pneumologie et Centre du Sommeil, Ho ^pital Neucha^telois, site de Pourtale s, Neucha ^tel, Switzerland; 19 MoMiLab Research Unit, IMT School for Advanced Studies, Lucca, Italy; 20The Netherlands Institute for Neuroscience, Amsterdam, The Netherlands; 21The Sense Innovation and Research Center, Lausanne and Sion, Switzerland; 22Department of Respiratory and Sleep Medicine, Angers University Hospital, Angers, France; and 23INSERM, CNRS, MITOVASC, Equipe CarME, SFR ICAT, University of Angers, Angers, France ORCID IDs: 0000-0002-7135-133X (G.S.); 0000-0002-5695-348X (M.S.-d.-l.-T.); 0000-0001-8382-4037 (M. Blanchard); 0000-0002-7809-4399 (N.A.M.); 0000-0002-2179-4650 (S.B.); 0000-0002-2340-8928 (F.B.); 0000-0002-4231-5102 (F. Gagnadoux); 0000-0002-3215-7788 (R.H.). treatment (3–5). Therefore, more specific events per hour (PWAD index) has been At a Glance Commentary biomarkers are needed to assess shown to vary according to the sleep stage, cardiovascular risk in OSA. with a lower frequency in deep sleep stage N3 Scientific Knowledge on the In contrast to the measurement of arm compared with stage N2 sleep, and a higher Subject: Obstructive sleep apnea pulse analysis, which requires an arm cuff, frequency in stage N1 and REM sleep (11, (OSA) severity has traditionally finger pulse wave analysis can be easily 12), suggesting a strong association with been assessed using the assessed using photoplethysomography autonomic nervous system reactivity. It is apnea–hypopnea index. This index (PPG), a signal derived from a pulse oximeter. interesting that a low PWAD index was does not take into account the PPG is a simple, low-cost, and noninvasive observed in conditions associated with complexity of respiratory events and technique that assesses oscillations in pulsatile increased cardiovascular risk, such as older may not be sufficient to assess finger blood volume on the basis of variations age, diabetes, and hypertension, and in OSA-associated cardiovascular risk of blood red and infrared light absorption (6). patients with an existing cardiovascular or to select patients likely to Extremities such as the fingers play an disease (12). respond to continuous positive important role in heat regulation by In OSA, there is a nightly occurrence of airway pressure treatment. modulating their highly variable skin blood hundreds of respiratory events, generating flow, which is predominantly regulated by blood pressure peaks, sympathetic and What This Study Adds to the sympathetic vasoconstrictor tone: Stimulation parasympathetic activations that could blunt Field: In patients with OSA, a low and blockade of vascular a receptors generate autonomic nervous system responsiveness, pulse wave amplitude drop index finger skin vasoconstriction and vasodilation, possibly through a “tolerance” of the was associated with a higher respectively (7–9). Grote and colleagues baroreflex, as observed by Parati and cardiovascular risk. showed that pulse wave amplitude drops colleagues (13). OSA has also been shown (PWADs; Figure 1) reflect transient to be associated with reduced endothelial vasoconstriction followed by a vasodilation function (impaired flow-mediated Obstructive sleep apnea (OSA) severity has that occurs in response to surges in vasodilation) (14). Considering that both traditionally been assessed using the sympathetic activity, which are then followed a poorly responsive autonomic system and apnea–hypopnea index (AHI), but this index by a compensatory parasympathetic response impaired endothelial function are associated does not take into account the complexity of (10). PWADs are typically observed with increased cardiovascular risk and that respiratory events and may not be sufficient concomitantly with cortical arousals that they also tend to decrease PWA variations, to assess OSA-associated cardiovascular risk occur spontaneously or after nocturnal events we hypothesized that patients with OSA who (1, 2) or to select patients likely to respond to such as sleep apneas/hypopneas and leg have a low PWAD index could be at higher continuous positive airway pressure (CPAP) movements (11). The number of PWAD risk of incident cardiovascular diseases. Author Contributions: Drafting/revision of the manuscript for content, including medical writing for content: G.S., M.S.-d.-l.-T., M. Blanchard, M. Berger, C.H., T.I., A.S.-d.-l.-T., N.A.M., A.W., G.H., S.B., F. Goupil, T.P., E.G.-L., A.Z., J.A., E.O., M.J.M., V.C.-N., C.E., V.B., S.V.D.B., J.H.-R., P.V., P.M.-V., J.V., G.B., M. Betta, F.S., F.B., F. Gagnadoux, and R.H. Verification of the underlying data: G.S., M. Blanchard, M.S.-d.-l.-T., and R.H. Analysis or interpretation of data: G.S., M. Blanchard, M. Berger, C.H., T.I., F. Gagnadoux, and R.H. Study concept or design: M.S.-d.-l.-T., F.B., F. Gagnadoux, and R.H. Acquisition of funding resources: F.B. and R.H. M.S.-d.-l.-T., J.H.-R., P.V., P.M.-V., J.V., F.B., F. Gagnadoux, and R.H. All had a major role in the acquisition of data. Correspondence and requests for reprints should be addressed to Raphael Heinzer, M.D., M.P.H., Center for Investigation and Research in Sleep, Lausanne University Hospital and University of Lausanne, Rue de Bugnon 46, 1011 Lausanne, Switzerland. E-mail: raphael.heinzer@chuv.ch. This article has a related editorial. This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org. nchez-de-la-Torre, Blanchard, et al.: Association of PWADs and Cardiovascular Risk in OSA Solelhac, Sa 1621 ORIGINAL ARTICLE Figure 1. PPG and PWA assessment. PPG = photoplethysmography; PWA = pulse wave amplitude; PWAD = pulse wave amplitude drop. Therefore, this study evaluated the French administrative healthcare database PWADs with an amplitude .30% from association between PWAD index and (SNDS) (18, 19). All overt cardiovascular baseline (see the online supplement) and a incident or recurrent cardiovascular events in disease–free patients investigated for OSA duration of more than four heartbeats were three prospective cohorts: HypnoLaus, using in-lab PSG or type 3 polygraphy identified (Figure 1). The total number of ISAACC, and the Pays de la Loire Sleep between May 15, 2007 and December 31, PWADs during the whole night was Cohort [PLSC]). A secondary objective was to 2018, who had available SNDS data were averaged per hour of sleep as the PWAD assess the response to CPAP according to the included in the study (N = 6,367). index (number of .30% drops in PWA PWAD index. Some of these data have been per hour). previously reported in an abstract form (15, 16). ISAACC Cohort The ISAACC cohort study is a Spanish multicenter, prospective, randomized Cardiovascular Endpoint Methods controlled trial (ClinicalTrials.gov identifier In the HypnoLaus cohort, the composite NCT01335087) including adults admitted for incident cardiovascular endpoint included Data from three independent prospective an acute coronary syndrome event who had fatal cardiovascular events (death from cohorts were used: the HypnoLaus an Epworth Sleepiness Scale score of 10 or myocardial infarction or stroke) and nonfatal population-based cohort, the PLSC clinic- less (20). Patients with OSA were randomly cardiovascular events (myocardial infarction, based cohort, and the ISAACC cardiovascular assigned (1:1) to CPAP treatment plus usual stroke, transient ischemic attack, and cohort. (For a detailed description of these care (CPAP group) or usual care alone. coronary heart disease) (24). In PSLC, a cohorts, see the online supplement.) A group of patients with acute coronary major adverse cardiovascular event (MACE), syndrome but without OSA was included the primary composite outcome of the study, HypnoLaus Cohort as a reference group. was defined using SNDS data as fatal events The HypnoLaus sleep cohort study was (all-cause death) and nonfatal cardiovascular designed to assess the prevalence and OSA events (myocardial infarction, stroke, correlates of sleep-disordered breathing in a OSA was defined as an AHI of >15 or more transient ischemic attack, coronary heart general unselected middle-to-older-age events per hour, as assessed by PSG using the disease, and hospitalization for heart failure). population of Lausanne, Switzerland (17). 2012 and 2007 American Academy of Sleep Participants were followed from their first Briefly, 2,162 participants underwent a Medicine (AASM) criteria (21) in the diagnostic sleep study to the end of complete clinical assessment and overnight HypnoLaus study, by polygraphy using the December 2019, or the occurrence of a unattended full polysomnography (PSG) 2007 AASM criteria (22) in the ISAACC primary outcome event, whichever occurred between 2009 and 2013, followed by clinical study, and by polygraphy or PSG using the first (18). follow-up. 2012 AASM criteria (21) in the PLSC study. In the ISAACC cohort, the composite cardiovascular endpoint included fatal PLSC PWA Assessment cardiovascular events (death from This study used data collected by the French In all three cohorts, PWADs on the PPG myocardial infarction or stroke) or nonfatal clinic-based multicenter longitudinal study, signal of the pulse oximeter were identified cardiovascular events (myocardial infarction, which was further linked with data from the using the same validated algorithm (23). stroke, transient ischemic attack, 1622 American Journal of Respiratory and Critical Care Medicine Volume 207 Number 12 | June 15 2023 ORIGINAL ARTICLE hospitalization for heart failure, coronary previously, patients using CPAP at least the low-PWAD and high-PWAD groups heart disease, and atrial fibrillation). 4 hours per night during the entire follow-up (P = 0.575) (Figure 3; Table E1). OSA1 period were assigned to the CPAP-adherent participants in PWAD index quartiles 1 and Statistical Analysis group (25). Those who stopped the use of 2 (Q1 and Q2) had a significantly higher risk Baseline characteristics of participants with CPAP or used the device for less than 4 hours of incident cardiovascular events compared and without cardiovascular events and per night constituted the nonadherent group. with those from quartile 4 (Q4) (Q4 vs. Q1: groups with (OSA1, AHI >15/h) or without Subgroups were defined according to CPAP adjusted HR, 3.09 [95% CI, 1.00–9.55], (OSA–, AHI ,15/h) OSA were compared usage (adherent vs. nonadherent) and PWAD P = 0.049; Q4 vs. Q2: HR, 3.43 [95% CI, using Student’s t test or chi-square test as (high vs low) index. The same analysis was 1.05–11.1], P = 0.041). There was no appropriate. performed in the ISAACC cohort, with significant difference between quartiles in Interaction between OSA and PWAD subgroups defined according to CPAP usage OSA2 participants (Figure 4). When used as (high or low) was tested. Four groups were (CPAP treatment vs. usual care) and PWAD a continuous variable, PWAD index (110 created in the HypnoLaus cohort and the (high vs low) index. events/h) was significantly associated with PLSC on the basis of the presence or absence Sensitivity analyses were conducted lower risk of incident cardiovascular events of OSA (OSA1 and OSA–, respectively) and using the 2007 AASM criteria (22) for AHI in participants with an AHI >15/h but not in the median PWAD index (high or low); that in the HypnoLaus cohort and different AHI those with an AHI ,15/h (HR, 0.85 [95% is, low PWAD and OSA1, high PWAD and cutoffs (10/h, 20/h, and 30/h) for the CI, 0.73–0.99], P = 0.031; and HR, 0.94 [95% OSA1, low PWAD and OSA–, and high HypnoLaus cohort and PLSC. Additional CI, 0.78–1.13], P = 0.495, respectively). PWAD and OSA–. sensitivity analyses were performed to better Adjusted Cox models showed a significant For the ISAACC cohort, five groups harmonize the MACE outcome by removing interaction between PWAD and AHI as were defined to take into account the heart failure hospitalizations in PLSC and continuous variables (P = 0.016) but not treatment intervention: OSA–; OSA1 (usual atrial fibrillation and heart failure when AHI was dichotomized (>15/h and care) and low PWAD, OSA1 (usual care) hospitalizations in the ISAACC cohort. (For ,15/h, respectively, P = 0.190). and high PWAD, OSA1 (CPAP)/low additional methodological details, see the A sensitivity analysis showed the same PWAD, and OSA1 (CPAP)/high PWAD. online supplement.) differences between the low-PWAD and To determine the association between OSA1 and high-PWAD and OSA1 groups PWAD and the occurrence of a for other AHI cutoff values (see Table E2). cardiovascular composite endpoint, Results A significant interaction between a multivariable-adjusted Cox regressions were continuous AHI defined using the 2007 performed in each cohort according to the HypnoLaus Cohort AASM criteria and a continuous PWAD different population subgroups defined Of the 2,162 participants initially included index was also found (P = 0.007). Slight earlier and according to PWAD quartiles. in the HypnoLaus cohort, 1,941 variations in associations were seen at For each cohort, Cox models were adjusted with complete PSG and PWAD analysis different AHI cutoff values when 2007 for age, sex, body mass index, alcohol intake, who had no cardiovascular events at AASM criteria were used (Table E2). smoking, diabetes, hypertension, lipid- baseline, were included in this study lowering drugs, and vasodilators, plus CPAP (Table 1; see Figure E2). In this study, 3.9% PLSC usage (for PLSC only). Adjustment factors of participants experienced a cardiovascular Among 9,790 patients from the PLSC, 1,995 representing clinically recognized or event over a (mean 6 SD) follow-up period had no available SNDS data, 1,428 were historical cardiovascular risk factors were of 49.2 6 12.1 months. The median excluded because of prior cardiovascular used. The directed acyclic graph describing (interquartile range [IQR]) PWAD index disease and 6,367 were included in the the relationships between PWAD, risk of was 51.8 (37.7–65.5) per hour (Figure 2). present study (Table 1; see Figure E3). In the cardiovascular events, and adjustment factors There was no difference in the incidence PLSC, 10.5% experienced a MACE over is shown in the online supplement (see of cardiovascular events between patients (mean 6 SD) follow-up period of Figure E1 in the online supplement). Results with and without OSA (AHI >15/h vs. 72.2 6 35.4 months, including 256 all-cause are expressed as hazard ratios (HRs) with ,15/h) in an adjusted Cox model. Results deaths and 413 nonfatal cardiovascular 95% confidence interval (95% CI) values. were similar when other AHI cutoff values, events. The median (IQR) PWAD index was The association between PWAD index as a continuous AHI, or AASM 2007 AHI criteria 46.9 (33.6–60.3) per hour (Figure 2). continuous variable with cardiovascular (22) were used instead of the AASM 2012 The highest MACE risk was seen in the events was analyzed in Cox models adjusted definition of AHI (21) (see Table E1). OSA1/low-PWAD group (adjusted HR, for the same confounding factors among The risk of incident cardiovascular 1.36 [95% CI, 1.13–1.63], when compared groups with an AHI >15 and ,15/h in the events was significantly higher in the low- with the OSA1/high-PWAD group; HR, three cohorts. The model assumptions were PWAD and OSA1 group compared with 1.44 [95% CI, 1.06–1.94], when compared investigated using a Schoenfeld’s test. the high-PWAD and OSA1 group (HR, 2.16 with the OSA–/high-PWAD group; and HR, A secondary analysis was conducted to [95% CI, 1.07–4.34]), the high-PWAD and 1.10 [95% CI, 0.88–1.38], when compared evaluate CPAP response according to the OSA– group (HR, 2.35 [95% CI, 1.12–4.93]), with the OSA–/low-PWAD group) (Figure 5; PWAD index by including patients with and the low-PWAD and OSA– group see Table E3). Similar findings were observed OSA from PLSC who had started CPAP (HR, 1.90 [95% CI, 1.02–3.54]). In OSA– for other AHI threshold values (Table E3). In treatment at home between May 15, 2007, participants, no significant difference in the OSA1 group, patients in the lowest and December 31, 2018. As described cardiovascular event rate was found between PWAD index quartile (Q1) had a nchez-de-la-Torre, Blanchard, et al.: Association of PWADs and Cardiovascular Risk in OSA Solelhac, Sa 1623 1624 Table 1. Baseline Demographic Characteristics of the HypnoLaus Cohort, Pays de la Loire Sleep Cohort, and ISAACC Cohort in Patient Subgroups that Are Based on the Presence or Absence of Recurrent Cardiovascular Events HypnoLaus ISAACC PLSC No incident Incident No recurrent Recurrent No incident Incident CVEs CVEs CVEs CVEs MACEs MACEs Variable (n = 1,866) (n = 75) P Value (n = 575) (n = 117) P Value (n = 5,698) (n = 669) P Value Age, yr 57 [49–68] 68 [55–74] ,0.001 59.0 6 10.1 61.2 6 10.5 0.040 58 [48–68] 70 [61–78] ,0.001 Female, n (%) 1,000 (53.6) 28 (37.3) 0.006 93 (16.2) 21 (17.9) 0.637 2,301 (40.4) 181 (27.1) ,0.001 BMI, kg/m2 25.5 [23.0–28.4] 27.2 0.002 28.4 [25.9–31.2] 28.7 0.328 29.0 [25.3–34.0] 30.3 ,0.001 [24.3–29.9] [26.0–32.0] [26.9–34.5] SBP, mm Hg 125 6 17 134 6 21 ,0.001 120 [111–132] 122 0.873 128 [120–135] 130 [125–140] ,0.001 [110–131] DBP, mm Hg 78 6 10 78 6 13 0.559 72 [65–80] 71.0 0.524 77 [73–81] 77 [72–83] 0.111 [55–79] Education, yr 13 [12–16] 12 [11–14] 0.028 — — — — — — Medication use, n (%) Vasodilators* 406 (21.8) 36 (48.0) ,0.001 240 (41.7) 69 (59.0) ,0.001 510 (9.0) 155 (23.2) ,0.001 Antidiabetics 81 (4.3) 8 (10.7) 0.010 106 (18.4) 40 (34.2) ,0.001 544 (9.5) 149 (22.3) ,0.001 Lipid-lowering agents 341 (18.3) 25 (33.3) 0.001 196 (34.1) 62 (53.0) ,0.001 968 (17.0) 269 (40.2) 0.007 Comorbidities, n (%) Hypertension 706 (37.9) 53 (70.7) ,0.001 291 (50.6) 75 (64.1) 0.008 1,297 (22.8) 290 (43.3) ,0.001 Diabetes 159 (8.5) 14 (18.7) 0.003 126 (21.9) 46 (39.3) ,0.001 544 (9.5) 149 (22.3) ,0.001 Dyslipidemia 480 (25.8) 29 (38.7) 0.013 325 (56.5) 76 (65.0) 0.092 968 (17.0) 269 (40.2) 0.007 Smoking, n (%) — — 0.001 — — 0.816 — — ,0.001 Current smoker 325 (17.6) 24 (32.4) — 294 (51.1) 59 (50.4) — 1,365 (24.0) 144 (21.5) — Ex-smoker 731 (39.6) 32 (43.2) — 153 (26.6) 29 (24.8) — 1,829 (32.0) 269 (40.2) — Alcohol consumption — — 0.718 — — 0.146 — — ,0.001 Alcohol units, standard 4 [1–9] 4 [0–11] — — — — 0 [0–7] 7 [0–14] — drinks,† wk Cessation, n (%) — — — 4 (0.7) 3 (2.6) — — — — Current consumption, n (%) — — — 126 (21.9) 22 (18.8) — — — — Sleep parameters ESS score 6 [3–9] 5 [2–8] 0.095 4.0 [3.0–6.0] 5.0 0.660 10 [6–14] 9 [6–13] ,0.001 [43.0–7.0] ‡ TST, min 402 6 72 395 6 90 0.356 — — — 421 [372–463] 390 0.193 [348–447]‡ Sleep efficiency, % 88.2 [80.3–92.8] 83.4 0.002 — — — 87 [79–92]‡ 81 [73–89]‡ 0.051 [75.2–90.4] ‡ N1 sleep, % 10.1 [7.2–14.4] 11.3 0.086 — — — 6.3 [3.4–10.3] 7.9 ,0.001 [7.2–16.7] [4.0–13.5]‡ N2 sleep, % 46.4 6 10.3 50.0 6 11.6 0.003 — — — 50.5 [43.2–57.7]‡ 50.0 0.032 [42.0–58.3]‡ ‡ N3 sleep, % 19.9 6 8.4 16.3 6 8.6 ,0.001 — — — 21.8 [16.5–28.2] 21.1 0.054 [15.0–27.8]‡ REM sleep, % 21.9 6 6.1 19.6 6 7.1 0.002 — — — 19.3 [14.3–23.5]‡ 17.9 0.125 [12.0–22.9]‡ (Continued) ORIGINAL ARTICLE American Journal of Respiratory and Critical Care Medicine Volume 207 Number 12 | June 15 2023 ORIGINAL ARTICLE significantly higher incidence of MACEs Definition of abbreviations: AHI = apnea–hypopnea index; BMI = body mass index; CVEs = cardiovascular events; DBP = diastolic blood pressure; ESS = Epworth Sleepiness Scale; MACEs = major adverse cardiovascular events; N1 = sleep stage 1; N2 = sleep stage 2; N3 = sleep stage 3; PLSC = Pays de la Loire Sleep Cohort; OSA = obstructive sleep apnea P Value 0.146 ,0.001 ,0.001 ,0.001 ,0.001 ,0.001 compared with those in the highest quartile study; angiotensin receptor blockers, angiotensin-converting enzyme inhibitors, and b-blockers for the ISAACC study; and b blockers or calcium channel blockers for the PLSC *Vasodilators: a-adrenoreceptor antagonists, antiadrenergic agents, angiotensin receptor blockers, angiotensin-converting enzyme inhibitors, and b-blockers for the HypnoLaus Data are presented as number of patients (%), median [25th–75th percentile], or mean 6 SD. Data were analyzed using Pearson’s chi-square test, independent t test, or Mann- (defined as AHI >15 events/h); PWAD = pulse wave amplitude drop; SBP = systolic blood pressure; SpO2 = oxygen saturation as measured by pulse oximetry; TST = total sleep (Q4): For Q4 vs. Q1, HR, 1.55 (95% CI, 1.18–2.04), P = 0.002. There was no significant difference between quartiles in 3.4 [0.5–13.8] 29 [17–40]‡ [13.0–43.0] [92.1–94.6] [29.6–56.5] 474 (70.9) the OSA2 group (Figure 4). (n = 669) Incident MACEs Similar to the HypnoLaus findings, the 27.5 93.3 41.9 continuous PWAD index (110 events/h) PLSC was significantly associated with a lower risk of incident cardiovascular events in participants with an AHI >15/h but not 94.1 [92.8–95.3] 47.4 [34.2–60.7] 17.6 [7.5–32.5] in those with an AHI ,15/h (HR, 0.91 0.8 [0.1–5.2] No incident 3,190 (56.0) 24 [15–34]‡ (n = 5,698) MACEs [95% CI, 0.86–0.96], P , 0.001; and HR, 0.97 [95% CI, 0.89–1.05], P = 0.460, respectively). The interaction between continuous PWAD index and continuous or dichotomized AHI (>15/h and ,15/h) in the adjusted Cox P Value 0.005 0.024 0.007 0.091 0.115 model did not reach statistical significance (Ps = 0.068 and 0.706, respectively). — Similar findings were also observed when the analysis was restricted to nonfatal [19.1–44.5] [91.4–94.2] [22.1–51.6] Recurrent 101 (86.3) [2.0–82.5] cardiovascular events alone and when (n = 117) CVEs 28.4 11.2 93.0 34.5 excluding heart failure hospitalizations — ISAACC (see Figures E4 and E5). Patients with OSA who had started CPAP between May 15, 2007, and December 22.6 [14.0–40.8] 93.1 [92.0–94.2] 39.8 [25.5–54.2] 7.0 [0.7–35.9] No recurrent 31, 2018, (n = 3,669) were divided into four 427 (74.3) (n = 575) groups according to treatment adherence CVEs (>4 h per night and ,4 h or treatment stop) — and high versus low PWAD index values (see Table E4). As shown in Figure 6, CPAP adherence versus nonadherence was Data were available for 3,374 patients who were investigated by polysomnography. P Value 0.026 ,0.001 ,0.001 ,0.001 ,0.001 ,0.001 associated with a significant reduction in incident MACEs in the high-PWAD group time; T90 = percentage of total sleep time with oxygen saturation below 90%. (HR, 0.74 [0.55–0.98], P = 0.036) but not in the low-PWAD group (HR, 0.91 [0.72–1.15], P = 0.431), even though the interaction [15.9–30.3] [92.0–94.4] [26.4–58.9] 1.1 [0–8.6] [6.9–31.6] 43 (57.3) Incident (n = 75) CVEs between PWAD index and CPAP adherence 20.5 17.5 93.5 44.9 HypnoLaus was not significant (P = 0.229). ISAACC Of the 1,773 ISAACC participants, a subset 18.9 [13.9–26.0] 94.3 [93.2–95.4] 52.2 [38.2–65.7] 9.6 [4.0–19.9] No incident of 692 studied in the last 5 years of the study 0.1 [0–1.7] (n = 1,866) 647 (34.7) and for whom PPG signal data were available CVEs Whitney pairwise comparisons, as appropriate. was included in the present analysis (Table 1; A standard drink contains 10 g of alcohol. see Figure E6). (For characteristics of included and excluded ISAACC participants, see Table E5). Overall, 16.9% of the ISAACC cohort experienced a cardiovascular event over a mean follow-up period of Arousal index, events/h 24.3 6 15.8 months. The median (IQR) PWAD index, events/h Table 1. (Continued) PWAD index was 39.2 (25.0–53.5) per hour (Figure 2). Mean SpO2, % Patients in the low-PWAD and OSA1/ AHI, events/h OSA, n (%) usual care group had a higher risk of recurrent cardiovascular events compared Variable T90, % with the OSA– group (HR, 2.03 [95% CI, study. 1.08–3.81]); this difference was not found in other OSA groups (treated patients with † ‡ nchez-de-la-Torre, Blanchard, et al.: Association of PWADs and Cardiovascular Risk in OSA Solelhac, Sa 1625 ORIGINAL ARTICLE Figure 2. PWAD index distributions in the three different cohorts. Values are presented as hazard ratios and 95% confidence intervals. PLSC = Pays de la Loire Sleep Cohort; PWAD = pulse wave amplitude drop. OSA who had a low PWAD index: HR, 1.73 also no significant difference between treatment) and in those with AHI ,15/h [95% CI, 0.90–3.36] compared with OSA–; PWAD index quartiles in OSA1 and OSA2 (HR, 0.94 [95% CI, 0.81–1.10], P = 0.438; and and high-PWAD and OSA1/usual care: HR, groups (Figure 4). HR, 1.14 [95% CI, 0.80–1.63], P = 0.458, 1.56 [95% CI, 0.78–3.10] compared with Sensitivity analyses showed that a respectively). Adjusted Cox models showed a OSA–) or when atrial fibrillation or heart continuous PWAD index (110 events/h) significant interaction between continuous failure hospitalizations were removed from was not significantly associated with lower AHI and continuous PWAD index (P = 0.018) the cardiovascular composite endpoint risk of recurrent cardiovascular events in but not when AHI was dichotomized in (Figure 7; see Figures E7 and E8). There was patients with an AHI >15/h (without CPAP >15/h and ,15/h (P = 0.713). CPAP Figure 3. HypnoLaus study: adjusted Cox regression curves showing the incidence of composite cardiovascular events on the basis of (A) the presence or absence of obstructive sleep apnea (OSA1 and OSA– , respectively) and (B) OSA1 or OSA– and high or low pulse wave amplitude drop (PWAD) index. Models were adjusted for age (continuous), sex (categorical), body mass index (continuous), weekly alcohol consumption (continuous), smoking (continuous), diabetes (categorical), hypertension (categorical), lipid-lowering drugs (categorical), and vasodilators (categorical). Group description (n, median PWAD index value as n/h [95% confidence interval]): OSA– (1,251, 51.4 [37.2–64.5]), OSA1 (690, 52.4 [38.9–67.6]), OSA–/high PWAD index (629, 64.3 [57.9–73.1]), OSA–/low PWAD index (618, 37.1 [26.9–44.5]), OSA1/high PWAD index (359, 66.3 [58.4–75.2]), and OSA1/low PWAD index (323, 38.4 [28.4–44.9]). aP = 0.031 for OSA1/low PWAD index versus OSA1/high PWAD index. bP = 0.045 for OSA1/low PWAD index versus OSA–/low PWAD index. cP = 0.024 for OSA1/low PWAD index versus OSA–/high PWAD index. *P , 0.05, statistically significant. NS = not statistically significant. 1626 American Journal of Respiratory and Critical Care Medicine Volume 207 Number 12 | June 15 2023 ORIGINAL ARTICLE A No OSA OSA * Q1 * Q2 Q3 Q1 Q2 Q4 Q3 Q4 B No OSA OSA * Q1 Q2 Q3 Q1 Q4 Q2 Q3 Q4 C No OSA OSA Q1 Q2 Q2 Q4 Q3 Q3 Q1 Q4 Figure 4. Adjusted Cox regression curves showing the incidence or recurrence of cardiovascular events on the basis of pulse wave amplitude drop index quartiles (Qs) in the presence (right) or absence (left) of obstructive sleep apnea (OSA) in the three cohorts: *P , 0.05 and **P , 0.01, statistically significant. Models were adjusted according to the same variables as described in Figure 3 for HypnoLaus, Figure 5 for Pays de la Loire Sleep Cohort (PLSC), and Figure 7 for ISAACC. Values for each cohort are hazard ratios and 95% confidence intervals (HRs [95% CIs]). (A) HRs [95% CIs] for the HypnLaus cohort. No OSA: Q4 versus Q1, 1.27 [0.41–3.88], P = 0.677; Q4 versus Q2, 1.32 [0.43–4.07], P = 0.629; Q4 versus Q3, 1.004 [0.30–3.55], P = 0.994. OSA: Q4 versus Q1, 3.09 [1.00–9.55], P = 0.049; Q4 versus Q2, 3.43 [1.05–11.1], P = 0.041; and Q4 versus Q3, 2.50 [0.69–9.12], P = 0.165. No OSA quartiles (events/h): Q1, 0.23–37.11; Q2: 37.11–51.29; Q3: 51.29–64.52; and Q4: 64.52–119.14. OSA quartiles (events/h): Q1: 0.14–38.80; Q2: 38.80–52.36; Q3: 52.36–67.26; and Q4: 67.26–111.76. (B) HRs [95% CIs] for the PLSC. No OSA: Q4 versus Q1, 1.00 [0.65–1.53], P = 0.984; Q4 versus Q2, 1.10 [0.72–1.69], P = 0.652; and Q4 versus Q3, 0.80 [0.50–1.28], P = 0.356. OSA: Q4 versus Q1, 1.55 [1.18–2.04], P = 0.002; Q4 versus Q2, 1.30 [0.98–1.72], P = 0.069; and Q4 versus Q3, 1.11 [0.83, 1.49], P = 0.467. No OSA quartiles (event/h): Q1, 0.50–31.60; Q2, 31.60–44.08; Q3, 44.08–56.44; and Q4, 56.44–118.18. OSA quartiles (event/h): Q1, 2.20–35.39; Q2, 35.39–49.34; Q3, 49.34–62.78; and Q4, 62.78–126.36. (C) HRs [95% CIs] for the ISAACC cohort. No OSA: Q4 versus Q1, 0.50 [0.06–4.02], P = 0.512; Q4 versus Q2, 1.34 [0.24–7.25], P = 0.736; and Q4 versus Q3, 1.29 [0.21–7.72], P = 0.780. OSA: Q4 versus Q1, 1.26 [0.58–2.97], P = 0.520; Q4 versus Q2, 1.00 [0.41–2.44], P = 0.992; Q4 versus Q3, 0.79 [0.32–1.98], P = 0.628. No OSA quartiles (event/h): Q1, 4.44–23.19; Q2, 23.19–35.79; Q3, 35.79–47.90; and Q4, 47.90–92.81. OSA quartiles (events/h): Q1, 1.70–24.76; Q2, 24.76–40.30; Q3, 40.30–53.08; and Q4, 53.08–118.10. nchez-de-la-Torre, Blanchard, et al.: Association of PWADs and Cardiovascular Risk in OSA Solelhac, Sa 1627 ORIGINAL ARTICLE Figure 5. Pays de la Loire Sleep Cohort: adjusted Cox regression curves showing the incidence of major adverse cardiovascular events on the basis of the presence or absence of obstructive sleep apnea (OSA1 and OSA–, respectively) and high or low pulse wave amplitude drop (PWAD) index. Models were adjusted for age (continuous), sex (categorical), body mass index (continuous), diabetes (categorical), hypertension (categorical), smoking status (categorical), type of sleep study (categorical), study site (categorical), b blocker/calcium channel blocker medications (categorical), and treatment status (categorical). Group description (n, median PWAD index value as n/h [95% confidence interval]): OSA–/high PWAD index (1,182, 58.5 [52.2–67.2]), OSA–/low PWAD index (1,522, 33.3 [24.2–40.0]), OSA1/high PWAD index (2,001, 61.1 [53.8–70.7]), and OSA1/low PWAD index (1,662, 33.9 [25.7–40.7]). aP = 0.001 for OSA1/low PWAD index versus OSA1/high PWAD index. b P = 0.410 for OSA1/low PWAD index versus OSA–/low PWAD index. cP = 0.019 for OSA1/low PWAD index versus OSA–/high PWAD index. *P , 0.05, statistically significant. treatment group (vs. usual care group) was outcome definitions suggest a good The statistical interaction found not associated with a reduction in incident reproducibility and generalizability of these between OSA and PWAD, as well as the cardiovascular events in the high-PWAD results. In addition, a subanalysis of PLSC presence of significant difference in group (HR, 1.16 [95% CI, 0.60–2.26], showed that a good adherence to CPAP in cardiovascular risk between high and low P = 0.662) or in the low-PWAD group (HR, patients with OSA with a high PWAD PWAD and between PWAD index quartiles 0.76 [95% CI, 0.41–1.39], P = 0.370). The index (i.e., without major alterations of exclusively in participants with OSA, interaction between PWAD index and CPAP autonomic or vascular function) was suggests that the PWAD index is influenced usage was not significant (P = 0.347). associated with a reduction in incident by OSA and could represent a specific Differences between OSA1 and OSA– cardiovascular events. marker of cardiovascular alterations caused participants from all three cohorts are by OSA. We can further hypothesize that summarized (see Table E6). Differences this marker could reflect the duration of between participant characteristics of Benefits of Adding PWAD Index to exposure to OSA, with a progressive the four groups in the HypnoLaus cohort the AHI blunting of autonomic nervous system and and PLSC (see Table E7) and the five In our analysis, the AHI alone at different vascular reactivity over time. This could groups in ISAACC (see Table E8) are also thresholds was not a reliable predictor of explain why only patients with OSA who provided. incident or recurrent cardiovascular events have high PWAD (i.e., those with preserved in any of the three cohorts analyzed. Several endothelial and autonomic nervous system prospective studies that selected participants reactivity) obtained some cardiovascular Discussion on the basis of the AHI have failed to protection from CPAP treatment in PLSC, demonstrate a preventive effect of CPAP on whereas those with low PWAD (i.e., with In this analysis of data from three recurrent cardiovascular events in patients greater and maybe irreversible impact of prospective cohorts, a low PWAD index in with OSA (3–5), probably because the AHI OSA on cardiovascular system) and patients patients with OSA was independently alone may not be sufficient to select patients from ISAACC who had established associated with a higher risk of incident who are likely to benefit from a reduction in cardiovascular disease did not. This cardiovascular events compared with cardiovascular risk during CPAP therapy. hypothesis requires confirmation in other participants who had OSA and a high Our results suggest that adding the PWAD prospective cohorts. PWAD index or those without OSA. The index to the AHI significantly improves the fact that similar findings were obtained in prediction of cardiovascular risk in patients Impact of OSA on PWAD: different populations (general, clinical, and with OSA and possibly allows the selection Physiological Hypothesis cardiovascular) and that studies that utilized of patients likely to benefit from treatment There are several potential explanations for slightly different cardiovascular endpoint with CPAP. the observed association between low 1628 American Journal of Respiratory and Critical Care Medicine Volume 207 Number 12 | June 15 2023 ORIGINAL ARTICLE Figure 6. Pays de la Loire Sleep Cohort: adjusted Cox regression curves showing the incidence of major adverse cardiovascular events on the basis of the use of (or not using) continuous positive airway pressure (CPAP) in high (left) or low (right) pulse wave amplitude drop (PWAD) index. Models were adjusted for age (continuous), sex (categorical), body mass index (continuous), diabetes (categorical), hypertension (categorical), smoking status (categorical), type of sleep study (categorical), study site (categorical), b blocker/calcium channel blocker medications (categorical), and treatment status (categorical). Group description (n, median PWAD index value as n/h [95% confidence interval]): high PWAD index-CPAP adherent (1,382, 60.7 [53.7–70.3]), high PWAD index-no CPAP adherent (454, 61.3 [52.9–69.6]), low PWAD index-CPAP adherent (1,308, 33.2 [24.1–40.8]), and low PWAD index-no CPAP adherent (525, 33.3 [25.8–39.9]). HR = hazard ratio. PWAD index and cardiovascular events in adequate autonomous nervous system PWAD index could suggest that finger patients with OSA. First, because drops in reactivity is needed for the cardiovascular arteries and capillaries have an impaired PWA occur concomitantly with EEG system to adapt to environmental events ability to rapidly redilate after activations (11) associated with sudden and stress, an impaired response could vasoconstriction. Impaired endothelial increases in heart rate and that they potentially increase cardiovascular risk. function (based on flow-mediated can be reproduced by the infusion of Lombardi and colleagues, who extensively vasodilation) has previously been described norepinephrine (10), we believe that studied autonomic nervous system in patients with OSA (14). Given that PWADs reflect sudden activations of the alterations in OSA, even proposed that impaired vascular (endothelial) function has sympathetic nervous system followed by a reduced baroreflex sensitivity could be a been associated with elevated cardiovascular compensatory parasympathetic response. A feature of OSA before the onset of risk, it is likely that lower PWAD index low rate of PWAD in OSA may, therefore, cardiovascular complications and that the would be associated with a higher incidence reflect a poorly reactive autonomic nervous chronic hypertensive state associated with of cardiovascular events (29). system. An association between OSA and OSA might be viewed as the result of Finally, patients with OSA have been impaired autonomous nervous system autonomic nervous system maladaptation shown to have increased sympathetic activity activity has been previously described and (26). Furthermore, in the Multicentre on the basis of microneurography, heart rate could be mediated through a decrease in Obstructive Sleep Apnea Interventional variability, and catecholamine levels (30–35). baroreceptor sensitivity: untreated OSA Cardiovascular study, patients treated with It could thus be hypothesized that patients with nightly occurrence of hundreds of therapeutic levels of CPAP showed a with OSA who have a low PWAD index respiratory events generating blood significant increase in baroreflex sensitivity have a constantly increased sympathetic pressure peaks, sympathetic and compared with patients randomized to tone with consistent vasoconstriction. This parasympathetic activations could induce subtherapeutic CPAP, suggesting that OSA vasoconstriction could generate a ceiling a “tolerance” of the baroreflex (13). This does impact baroreflex sensitivity (27, 28). effect, making further variations in finger tolerance effect could then decrease the The second hypothesis is that a low blood flow less likely. Constant sympathetic amplitude and the frequency of PWAD PWAD index may reflect impaired activation could then have deleterious effects because of blunted response of the endothelial function or arterial stiffness. on the cardiovascular system. These three autonomic system. The fact that a lower Because PWAD episodes are characterized hypotheses are not mutually exclusive, but PWAD index is associated with older age by vasoconstriction (drop in PPG signal) further physiological studies are needed and with higher blood pressure also followed by vasodilation when the PWA to determine the specific mechanisms supports this hypothesis (12). Because signal returns to normal values, a low linking low PWAD to OSA, increased nchez-de-la-Torre, Blanchard, et al.: Association of PWADs and Cardiovascular Risk in OSA Solelhac, Sa 1629 ORIGINAL ARTICLE Figure 7. ISAACC study: adjusted Cox regression curves showing incident or recurrent cardiovascular events based on low or high pulse wave amplitude drop (PWAD) index and treatment with continuous positive airway pressure (CPAP) or usual care for participants with obstructive sleep apnea (OSA) compared with the control group without OSA. Models were adjusted for age (continuous), sex (categorical), body mass index (continuous), alcohol (categorical), smoking (categorical), diabetes (categorical), hypertension (categorical), lipid-lowering drugs (categorical), and vasodilators (categorical). Group description (n, median PWAD index value as n/h [95% confidence interval]): no OSA (164, 34.0 [22.7–46.7]); OSA-CPAP/high PWAD index (151, 54.6 [47.3–63.0]); OSA-CPAP/low PWAD index (112, 25.5 [19.7–32.0]); OSA-usual care and high PWAD index (129, 53.1 [46.7–62.8]); and OSA-usual care and low PWAD index (135, 24.4 [18.5–31.0]). *P = 0.028 for OSA-usual care and low PWAD index versus no OSA; P = 0.110 for OSA-CPAP/high PWAD index versus no OSA; P = 0.102 for OSA-CPAP/low PWAD index group versus no OSA; P = 0.208 for OSA-usual care and high PWAD index group versus no OSA. cardiovascular risk, and the direction of Azarbarzin and colleagues recently that patients with OSA who have a low these associations. showed that the sleep apnea–specific pulse- PWAD index could correspond to the rate response predicted cardiovascular left part of the “U” shaped curve, with a morbidity and mortality in patients with blunted autonomic response (low pulse PWAD and Response to OSA, with a U-shaped relationship where rate response) contributing to increased CPAP Treatment both high and low pulse rate responses to cardiovascular risk. We could also Although patients with OSA who had low respiratory events were associated with hypothesize that, similar to patients with a PWAD appeared to be at higher risk for increased cardiovascular risk in the Sleep high heart rate response, patients with cardiovascular events, a subanalysis of PLSC Heart Heath study (36). In the cardiovascular OSA who have a high PWAD index can showed that good adherence to CPAP was cohort RICCADSA (Randomized respond to CPAP treatment in terms of associated with a decrease in incident Intervention with Continuous Positive cardiovascular risk reduction, because they cardiovascular events only in patients with Airway Pressure in coronary artery disease do not have irreversible alteration of OSA who had a high PWAD index. [CAD] and OSAtrial cohort), the same group endothelial function and the autonomic Although the interaction between PWAD found that CPAP only reduced the risk of nervous system. However, further and CPAP was not statistically significant, recurrent cardiovascular events in OSA prospective analysis of combinations of this may suggest that CPAP could prevent patients with OSA who had a higher pulse different cardiovascular risk markers such incident cardiovascular disease in OSA only rate response (37). as hypoxic burden, time spent with an in the presence of preserved autonomic PWAD and pulse rate response differ in oxygen saturation below 90%, respiratory nervous system and/or vascular reactivity. In several aspects: First, PWAD events likely event durations, and OSA-associated contrast, patients with OSA who have a low reflect both autonomic nervous system symptoms are needed before any PWAD index might have more advanced reactivity and vascular endothelial clinical recommendations can be cardiovascular disease that may not be responsiveness, whereas the pulse rate proposed (18, 38, 39). reversible with CPAP. The absence of a response reflects mainly autonomic nervous significant impact of CPAP on recurrent system reactivity. Second, the PWAD index Strengths and Limitations cardiovascular events in ISAACC may be takes into account all variations of the PPG The prospective design of the three cohorts, due to the type of patients included in this signal (vasoconstrictions occurring the adjudication of incident cardiovascular study with established cardiovascular disease spontaneously or after respiratory events), events by a panel of experts (in HypnoLaus (secondary prevention), the exclusion of whereas the pulse rate response was analyzed and ISAACC), and the replication of the sleepy patients, and the low use of CPAP exclusively in association with respiratory results despite differences in the types of (,3 h/night). events. Despite these differences, we suggest populations and outcome definitions are the 1630 American Journal of Respiratory and Critical Care Medicine Volume 207 Number 12 | June 15 2023 ORIGINAL ARTICLE main strengths of this analysis. There are, Finally, we recognize that the three higher PWAD index and who may have a however, limitations that need to be cohorts included in this analysis differ lower impact of OSA on their cardiovascular acknowledged. First, the underlying substantially not only in terms of population system, good adherence to CPAP treatment mechanisms of the association between the type but also in participant characteristics appears to have the potential to prevent PWAD index and cardiovascular risk are and recording methods (brands of recorders incident cardiovascular events. 䊏 still unknown, and physiological studies in and pulse oximeters) used. The three cohorts patients with OSA who have a high or low also used slightly different definitions for Author disclosures are available with the PWAD index are needed to address this their composite cardiovascular endpoint: text of this article at www.atsjournals.org. question. Second, it is not yet clear whether Hospitalization for heart failure was not the PWAD index can be used in patients available in the HypnoLaus cohort. The Acknowledgment: The authors thank Prof. Gerard Waeber, Prof. Vincent Mooser, with preexisting cardiovascular disorders, ISAACC cardiovascular endpoint included Prof. Mehdi Tafti, Daniela Andries, and Nadia such as atrial fibrillation, because these atrial fibrillation, whereas the other cohorts Tobback for their important contribution to the conditions may affect PWAD analysis. did not (results were nonsignificant without HypnoLaus and CoLaus Cohorts; the Lausanne Third, although the PWAD index seems to atrial fibrillation in ISAACC), and there was population who volunteered to participate in the contribute to cardiovascular risk prediction no adjudication of the primary outcome in CoLaus and HypnoLaus studies; and the whole team of CoLaus|PsyCoLaus. The authors also in association with AHI and, possibly, relate PLSC. However, sensitivity analysis of thank the sleep technicians Gianpaolo Lecciso, to the response to CPAP, this study focused ISAACC and PLSC with and without David Albir-Haba, and Brian Bradley for their on the PWAD index (events per hour), hospitalizations for heart failure showed important help with photoplethysmography with an arbitrary threshold of 30% drop. consistent results. Moreover, the fact that a analysis. English-language editing assistance However, other PPG signal–derived low PWAD index was associated with was provided by Nicola Ryan, independent medical writer, funded by Lausanne University parameters such as the duration or increased cardiovascular risk using Hospital and University of Lausanne, magnitude (area under the curve) of polygraphy or PSG, in younger and older Lausanne, Switzerland. The Pays de la Loire PWAD will need to be investigated further subjects, with shorter or longer follow-up, Sleep Cohort (PLSC) investigators thank the and may potentially improve the and in the general, clinical, and French National Health Insurance for giving performance of this signal as a predictor of cardiovascular samples is reassuring and access to the French administrative healthcare database; the Pays de la Loire cardiovascular risk in OSA. A recent study confirms the strength and generalizability of Respiratory Health Research Institute; promoter showed that the “pulse wave attenuation this marker. of the PLSC from which the data used for this index” was negatively correlated with In conclusion, these results suggest that study were obtained; Sandrine Kerbrat estimated cardiovascular risk on the adding the PWAD index to AHI provides a and Emmanuel Oger, EA 7449 European Society of Cardiology and better indication of increased cardiovascular [Pharmacoepidemiology and Health Services Research]; and Julien Godey, Laetitia Moreno European Society of Hypertension matrix risk and may reflect the specific impact of and Marion Vincent, sleep technicians in the (40), which seems to corroborate the results OSA on autonomic and/or vascular Department of Respiratory and Sleep Medicine presented here. reactivity. In patients with OSA who have a of Angers University Hospital. References tissue blood volume, following sympathetic blockade. Physiol Meas 2001;22:389–396. 1. Pevernagie DA, Gnidovec-Strazisar B, Grote L, Heinzer R, 8. Alian AA, Shelley KH. Photoplethysmography. Best Pract Res Clin McNicholas WT, Penzel T, et al. On the rise and fall of the apnea- Anaesthesiol 2014;28:395–406. hypopnea index: a historical review and critical appraisal. J Sleep Res 9. Colombo R, Marchi A, Borghi B, Fossali T, Rech R, Castelli A, et al. 2020;29:e13066. Pulse photoplethysmographic analysis estimates the sympathetic 2. Malhotra A, Ayappa I, Ayas N, Collop N, Kirsch D, McArdle N, et al. activity directed to heart and vessels. Anesthesiology 2015;123: Metrics of sleep apnea severity: beyond the apnea-hypopnea index. 336–345. Sleep 2021;44:zsab030. 10. Grote L, Zou D, Kraiczi H, Hedner J. Finger plethysmography – a method 3. McEvoy RD, Antic NA, Heeley E, Luo Y, Ou Q, Zhang X, et al.; SAVE for monitoring finger blood flow during sleep disordered breathing. Investigators and Coordinators. CPAP for prevention of cardiovascular Respir Physiol Neurobiol 2003;136:141–152. events in obstructive sleep apnea. N Engl J Med 2016;375: 11. Delessert A, Espa F, Rossetti A, Lavigne G, Tafti M, Heinzer R. Pulse 919–931. wave amplitude drops during sleep are reliable surrogate markers of 4. Peker Y, Glantz H, Eulenburg C, Wegscheider K, Herlitz J, Thunstro € m E. changes in cortical activity. Sleep 2010;33:1687–1692. Effect of positive airway pressure on cardiovascular outcomes in 12. Hirotsu C, Betta M, Bernardi G, Marques-Vidal P, Vollenweider P, coronary artery disease patients with nonsleepy obstructive sleep apnea. Waeber G, et al. Pulse wave amplitude drops during sleep: clinical The RICCADSA randomized controlled trial. Am J Respir Crit Care Med significance and characteristics in a general population sample. Sleep 2016;194:613–620. (Basel) 2020;43:zsz322. nchez-de-la-Torre M, Sa 5. Sa  nchez-de-la-Torre A, Bertran S, Abad J, Duran- 13. Parati G, Di Rienzo M, Bonsignore MR, Insalaco G, Marrone O, Cantolla J, Cabriada V, et al. Effect of obstructive sleep apnoea and its Castiglioni P, et al. Autonomic cardiac regulation in obstructive sleep treatment with continuous positive airway pressure on the prevalence of apnea syndrome: evidence from spontaneous baroreflex analysis cardiovascular events in patients with acute coronary syndrome during sleep. J Hypertens 1997;15:1621–1626. (ISAACC study): a randomised controlled trial. Lancet Respir Med 2019; 14. Kohler M, Craig S, Nicoll D, Leeson P, Davies RJ, Stradling JR. 8:359–367. Endothelial function and arterial stiffness in minimally symptomatic 6. Hertzman AB. The blood supply of various skin areas as estimated by the obstructive sleep apnea. Am J Respir Crit Care Med 2008;178: photoelectric plethysmograph. Am J Physiol 1938;124:328–340. 984–988. 7. Babchenko A, Davidson E, Ginosar Y, Kurz V, Faib I, Adler D, et al. 15. Heinzer R, Sa  nchez-De-La-Torre M, Solelhac G, Berger M, Hirotsu C, Photoplethysmographic measurement of changes in total and pulsatile Haba-Rubio J, et al. Pulse wave amplitude variations: a new biomarker nchez-de-la-Torre, Blanchard, et al.: Association of PWADs and Cardiovascular Risk in OSA Solelhac, Sa 1631 ORIGINAL ARTICLE of cardiovascular risk in patients with obstructive sleep apnea in 28. Bonsignore MR, Parati G, Insalaco G, Marrone O, Castiglioni P, HypnoLaus and ISAACC cohorts. Eur Respir J 2021;58:OA3011. Romano S, et al. Continuous positive airway pressure treatment 16. Solelhac G, Sanchez-de-la-Torre M, Berger M, Hirotsu C, Marchi N, improves baroreflex control of heart rate during sleep in severe Haba-Rubio J, et al. Le PWAD (pulse wave amplitude drops), un obstructive sleep apnea syndrome. Am J Respir Crit Care Med 2002; nouveau biomarqueur du risque cardiovasculaire lie  au SAOS dans les 166:279–286. cohortes prospectives HypnoLaus 1 ISAACC. Med du Sommeil 2022; 29. Rubinshtein R, Kuvin JT, Soffler M, Lennon RJ, Lavi S, Nelson RE, et al. 19:52–53. Assessment of endothelial function by non-invasive peripheral arterial 17. Heinzer R, Vat S, Marques-Vidal P, Marti-Soler H, Andries D, Tobback N, tonometry predicts late cardiovascular adverse events. Eur Heart J et al. Prevalence of sleep-disordered breathing in the general 2010;31:1142–1148. population: the HypnoLaus study. Lancet Respir Med 2015;3:310–318. 30. Dissanayake HU, Bin YS, Ucak S, de Chazal P, Sutherland K, Cistulli PA. 18. Trzepizur W, Blanchard M, Ganem T, Balusson F, Feuilloy M, Girault JM, Association between autonomic function and obstructive sleep apnea: et al. Sleep apnea–specific hypoxic burden, symptom subtypes, and a systematic review. Sleep Med Rev 2021;57:101470. risk of cardiovascular events and all-cause mortality. Am J Respir Crit 31. Somers VK, Dyken ME, Clary MP, Abboud FM. Sympathetic neural Care Med 2022;205:108–117. mechanisms in obstructive sleep apnea. J Clin Invest 1995;96: 19. Blanchard M, Gerve  s-Pinquie  C, Feuilloy M, Le Vaillant M, Trzepizur W, 1897–1904. Meslier N, et al. Association of nocturnal hypoxemia and pulse rate 32. Tamisier R, Tan CO, Pepin JL, Levy P, Taylor JA. Blood variability with incident atrial fibrillation in patients investigated for pressure increases in OSA due to maintained neurovascular obstructive sleep apnea. Ann Am Thorac Soc 2021;18:1043–1051. sympathetic transduction: impact of CPAP. Sleep (Basel) 2015;38: 20. Esquinas C, Sa  nchez-de-la Torre M, Aldoma  A, Flores M, Martınez M, 1973–1980. Barcelo  A, et al.; Spanish Sleep Network. Rationale and methodology 33. Solin P, Kaye DM, Little PJ, Bergin P, Richardson M, Naughton MT. of the impact of continuous positive airway pressure on patients with Impact of sleep apnea on sympathetic nervous system activity in heart ACS and nonsleepy OSA: the ISAACC Trial. Clin Cardiol 2013;36: failure. Chest 2003;123:1119–1126. 495–501. 34. Hilton MF, Chappell MJ, Bartlett WA, Malhotra A, Beattie JM, Cayton RM. 21. Berry RB, Budhiraja R, Gottlieb DJ, Gozal D, Iber C, Kapur VK, et al.; The sleep apnoea/hypopnoea syndrome depresses waking vagal American Academy of Sleep Medicine. Rules for scoring respiratory tone independent of sympathetic activation. Eur Respir J 2001;17: events in sleep: update of the 2007 AASM Manual for the Scoring of 1258–1266. Sleep and Associated Events. Deliberations of the Sleep Apnea 35. Narkiewicz K, Montano N, Cogliati C, van de Borne PJ, Dyken ME, Definitions Task Force of the American Academy of Sleep Medicine. Somers VK. Altered cardiovascular variability in obstructive sleep J Clin Sleep Med 2012;8:597–619. apnea. Circulation 1998;98:1071–1077. 22. Iber C, Ancoli-Israel S, Chesson A, Quan SF. The AASM manual for the 36. Azarbarzin A, Sands SA, Younes M, Taranto-Montemurro L, Sofer T, scoring of sleep and associated events: rules terminology and technical Vena D, et al. The sleep apnea-specific pulse-rate response predicts specifications. Westchester, IL: American Academy of Sleep Medicine; cardiovascular morbidity and mortality. Am J Respir Crit Care Med 2007. 2021;203:1546–1555. 23. Betta M, Handjaras G, Ricciardi E, Pietrini P, Haba-Rubio J, Siclari F, 37. Azarbarzin A, Zinchuk A, Wellman A, Labarca G, Vena D, Gell L, et al. et al. Quantifying peripheral sympathetic activations during sleep by Cardiovascular benefit of CPAP in adults with coronary artery disease means of an automatic method for pulse wave amplitude drop and OSA without excessive sleepiness. Am J Respir Crit Care Med detection. Sleep Med 2020;69:220–232. 2022;206:767–774. 24. Beuret H, Hausler N, Nanchen D, Me an M, Marques-Vidal P, 38. Azarbarzin A, Sands SA, Stone KL, Taranto-Montemurro L, Messineo L, Vaucher J. Comparison of Swiss and European risk algorithms for Terrill PI, et al. The hypoxic burden of sleep apnoea predicts cardiovascular prevention in Switzerland. Eur J Prev Cardiol 2021;28: cardiovascular disease-related mortality: the Osteoporotic Fractures in 204–210. Men Study and the Sleep Heart Health Study. Eur Heart J 2019;40: 25. Justeau G, Bailly S, Gerve s-Pinquie  C, Trzepizur W, Meslier N, Goupil F, 1149–1157. et al.; ERMES Study Group. Cancer risk in patients with sleep apnoea 39. Blanchard M, Gerve s-Pinquie  C, Feuilloy M, Le Vaillant M, Trzepizur W, following adherent 5-year CPAP therapy. Eur Respir J 2022;59: Meslier N, et al.; ERMES Study Group. Hypoxic burden and heart rate 2101935. variability predict stroke incidence in sleep apnoea. Eur Respir J 2021; 26. Lombardi C, Pengo MF, Parati G. Obstructive sleep apnea syndrome and 57:2004022. autonomic dysfunction. Auton Neurosci 2019;221:102563. 40. Strassberger C, Zou D, Penzel T, Fietze I, Hedner J, Ficker JH, et al. 27. Kohler M, Pepperell JC, Casadei B, Craig S, Crosthwaite N, Stradling JR, Beyond the AHI-pulse wave analysis during sleep for recognition of et al. CPAP and measures of cardiovascular risk in males with OSAS. cardiovascular risk in sleep apnea patients. J Sleep Res 2021;30: Eur Respir J 2008;32:1488–1496. e13364. 1632 American Journal of Respiratory and Critical Care Medicine Volume 207 Number 12 | June 15 2023