Journal of Applied Physiology

Patterned cardiovascular responses to sleep and nonrespiratory arousals in a porcine model

Sandrine H. Launois, Joseph H. Abraham, J. Woodrow Weiss, Debra A. Kirby


Patients with obstructive sleep apnea experience marked cardiovascular changes with apnea termination. Based on this observation, we hypothesized that sudden sleep disruption is accompanied by a specific, patterned hemodynamic response, similar to the cardiovascular defense reaction. To test this hypothesis, we recorded mean arterial blood pressure, heart rate, iliac blood flow and vascular resistance, and renal blood flow and vascular resistance in five pigs instrumented with chronic sleep electrodes. Cardiovascular parameters were recorded during quiet wakefulness, during non-rapid-eye-movement and rapid-eye-movement sleep, and during spontaneous and induced arousals. Iliac vasodilation (iliac vascular resistance decreased by −29.6 ± 4.1% of baseline) associated with renal vasoconstriction (renal vascular resistance increased by 10.3 ± 4.0%), tachycardia (heart rate increase: +23.8 ± 3.1%), and minimal changes in mean arterial blood pressure were the most common pattern of arousal response, but other hemodynamic patterns were observed. Similar findings were obtained in rapid-eye-movement sleep and for acoustic and tactile arousals. In conclusion, spontaneous and induced arousals from sleep may be associated with simultaneous visceral vasoconstriction and hindlimb vasodilation, but the response is variable.

  • sleep apnea
  • swine
  • defense reaction

in patients with obstructive sleep apnea (OSA), apnea termination is associated with substantial cardiovascular changes: tachycardia, increase in arterial blood pressure, and decrease in stroke volume (37). Several experimental studies have demonstrated the contribution of sudden sleep disruption to these cardiovascular changes (3, 11, 27, 28). Subsequently, some authors have regarded the cardiovascular response to obstructive apnea to be equivalent to a response to arousal (13, 27). However, the characterization of the cardiovascular response to arousal, in particular to nonrespiratory arousals, is still incomplete. We hypothesized that sudden sleep disruption triggers a complex hemodynamic reaction. Furthermore, we hypothesized that regional hemodynamic measurements would allow a better characterization of the cardiovascular reaction to arousal than would blood pressure and heart rate (HR) alone. Arterial blood pressure and HR are commonly monitored to study hemodynamic changes that accompany arousals (11, 27), but these two variables provide an incomplete picture of the hemodynamic response and may not always distinguish among different cardiovascular responses. Cardiac output and peripheral vascular resistance measurements can help clarify hemodynamic changes. However, in awake animal models, dissociated local vascular resistance responses have been reported (33, 38, 39), suggesting that regional flow measurement may yield a more complete assessment of hemodynamic responses. To test our hypotheses, we modified a recently developed porcine model to study the effects of sleep and sleep apnea on cardiovascular function (26). In the present report, we measured global and regional hemodynamic variables during sleep and during spontaneous (SPA) and induced arousals in chronically instrumented juvenile pigs.



Eight juvenile Yorkshire female pigs were included in the study. One pig did not sleep under experimental conditions, one pig was excluded from analysis after postmortem examination revealed that the instrumented kidney was necrotic, and one pig died of abdominal complications. We report the results in five pigs. Their mean initial body weight (BW) was 11.5 ± 0.5 kg, corresponding to ∼4 wk of age. The study period lasted 3–5 wk. Mean BW at the time of death was 19.9 ± 1.5 kg.

The protocol was approved by the Animal Care Committee of the West Roxbury Veterans Affairs Medical Center.


Abdominal instrumentation.

Sedation was obtained with ketamine (10 mg/kg) and xylazine sodium (2 mg/kg). After an intravenous catheter had been placed in an ear vein, anesthesia was induced with thiopental sodium (18 mg/kg iv), and the animal was intubated. Anesthesia was maintained with halothane or isoflurane in oxygen. A ventral midline laparotomy was performed by using sterile techniques. Iliac and renal arteries were exposed, with care taken not to damage the surrounding nerves. Ultrasonic flow probes (Transonic Systems, Ithaca, NY) were placed around a renal artery and the common trunk or a branch of an iliac artery. A catheter was inserted in the contralateral common iliac artery and advanced into the aorta. A second catheter was introduced in a branch of the iliac vein and advanced into the inferior vena cava. The flow probes and catheters were passed subcutaneously to exit from the back of the animal. Catheters were filled with saline and closed securely with an obturator.

Sleep electrode instrumentation.

Approximately 1 wk after the first surgical procedure, sleep monitoring electrodes were implanted, following a procedure described previously (26). Briefly, animals were anesthetized as described above, and electroencephalogram (EEG) and electrooculogram stainless-steel electrodes were implanted through the frontal sinus. Electromyogram (EMG) electrodes were placed in a neck muscle through a small skin incision. Electrodes were attached to a small nine-pin connector secured to the skull with methyl methacrylate bone cement (Surgical Simplex, Howmedica, Rutherford, NJ).



During the postsurgical recovery period, to minimize stress related to handling and study conditions, animals were brought from the housing facility to the laboratory and placed in an open-top stainless steel cage (0.90 × 0.90 × 1.20 m) lined with a thick rubber mat. They were kept in the recording cage for 1–2 h on alternate days. All parameters were recorded as for an experimental session, but sleep was not interrupted. As animals became accustomed to experimental conditions, they routinely assumed a recumbent position in <15 min. They then displayed repeated short sleep-wake cycles, typical of juvenile pigs (29). On alternate days, catheters were filled with heparin solution (100 U/ml), and rectal temperature and BW were measured. Animals received prophylactic antibiotics (cefazolin daily and wycillin weekly, im).


Vigilance state was determined by EEG, electrooculogram, and EMG signals and behavior (26, 29). Briefly, quiet wakefulness (QW) was characterized by high-frequency, low-amplitude EEG, slow and rapid eye movements, and high EMG tone. Non-rapid-eye-movement (NREM) sleep was characterized by low-frequency, high-amplitude EEG, absence of eye movement, and decreased EMG tone. Rapid-eye-movement (REM) sleep was characterized by high-frequency and low-voltage EEG, presence of rapid eye movements, and loss of tone on the neck EMG, with intense twitches of the nose and limbs. Arousal was defined as an abrupt increase in EEG frequency, lasting 3 s or more (8). Arousals were accompanied by behavioral changes of variable intensity. To quantify the intensity of arousal behaviors, we established a seven-point scale, based on observations gathered in the laboratory before the present study (Table1). With the exception of behavioralscore 6 (“startle”), items followed a consistent progressive pattern; for example, a head movement (behavioral score 3) was always preceded by a blink and eye opening (behavioral scores 1 and 2, respectively). HR was derived from surface electrocardiogram via a cardiotachometer. Arterial blood pressure was measured by connecting the indwelling arterial catheter to a pressure transducer referenced to heart level. Mean arterial blood pressure (MAP) was displayed on the chart recorder and used for analysis. Iliac and renal flow probes were connected to a dual-channel flow meter (Transonic Systems). Mean iliac and renal blood flow (IBF and RBF, respectively) were displayed on the chart recorder and used for analysis. IBF and RBF were normalized to BW (IBF/BW and RBF/BW, respectively) to reduce variability because of the animal’s growth during the study period. Regional vascular resistances [iliac and renal vascular resistance (IVR and RVR, respectively)] were calculated as the ratio of MAP to regional flow. HR, MAP, and regional flows were calibrated at the beginning and the end of each experimental session.

View this table:
Table 1.

Behavioral arousal score in juvenile pigs

Experimental protocol.

Experimental sessions began after the pig had recovered from sleep electrode placement (1–3 days). The animal was transported from the housing facility to the recording laboratory on study days and placed in the recording cage. The animal spontaneously laid down prone or on its side 10–15 min after being placed in the cage. The study started after the pig had been supine and quiet for 10 min. Hemodynamic and sleep parameters were continuously displayed on a chart recorder at a paper speed of 1 mm/s for subsequent analysis. During the first two sessions, the animal slept undisturbed. During the following sessions, arousals were provoked in NREM and REM sleep by using an unquantified acoustic stimulus [loud metallic noise; acoustic arousal (AA)] or a tactile stimulus [1 ml of icy mist on the ear or touch on the back; tactile arousal (TA)]. Acoustic and tactile stimuli were applied after at least 1 min of sleep had been observed. SPA occurred during all sessions. All events, interventions, and postural changes were noted on the chart by one of the investigators. The recording sessions took place between 0900 and 1400. Experimental sessions lasted 1.5–3 h. Each animal was studied on several occasions.


At the end of the study, animals were killed with a lethal injection of thiopental sodium. The absence of renal or iliac iatrogenic stenosis was ascertained by postmortem inspection.

Data Analysis

Steady-state sleep and wakefulness periods.

On the basis of sleep parameters and animal behavior, 1- to 2-min segments of QW and steady-state NREM and REM sleep were identified. For REM episodes, the analysis excluded the initial, transitional segment during which MAP reached a new baseline. Episodes of NREM and REM sleep shorter than 1 min were not included in the analysis. Measurements were taken every 5 s, and a mean value was obtained for each segment. If no artifact was present, each episode of QW, NREM, and REM recorded during the experimental sessions was included in the analysis. Approximately 15% of the segments were discarded because of artifacts or poor signal quality. For each animal, mean values were obtained for each vigilance state. Between-state comparisons were performed by using a one-way ANOVA and post hoc analysis.

SPA and provoked arousals.

Arousals were identified by using EEG signals. To be considered for analysis, arousals had to be preceded by at least 1 min of NREM or REM sleep. SPA occurring after <1 min of sleep were not analyzed. Immediately after EEG changes, measurements were made every second for 15 s after the arousal. Variations in hemodynamic parameters were expressed as a percentage of the baseline value, i.e., the mean value for the steady-state NREM or REM segment preceding the arousal. Approximately 30% of the arousals were not analyzed because of artifacts, poor quality signal, or insufficient chart annotation. We analyzed 69 arousals: 42 SPA (31 in NREM, 11 in REM), 11 AA (7 in NREM, 4 in REM), and 16 TA (10 in NREM, 6 in REM). For each animal, we calculated a mean value of each hemodynamic parameter for each arousal condition (stimulus type, sleep stage, behavioral score). Two group comparisons of cardiovascular response to arousal were performed by using a two-tailed Student’s t-test.

Statistical analysis was performed with Statview 4.5 for Macintosh. Results are reported as means ± SE.


Effect of Sleep on Hemodynamic Variables

During QW, MAP was 82.1 ± 5.0 mmHg, HR was 148.1 ± 5.5 beats/min, RBF/BW was 5.8 ± 0.6 ml ⋅ min−1 ⋅ kg−1, and IBF/BW was 6.8 ± 1.2 ml ⋅ min−1 ⋅ kg−1. Calculated RVR and IVR were 16.4 ± 3.2 and 15.9 ± 4.9 mmHg ⋅ ml−1 ⋅ min ⋅ kg, respectively.

Comparison of hemodynamic variables in steady-state QW and NREM sleep did not reveal any significant changes in MAP or HR (Table2). With REM onset, arterial blood pressure consistently decreased to reach a new baseline, and during stable REM sleep, MAP was 16.4 ± 1.2% lower than in NREM (P < 0.01), whereas HR increased by 12.2 ± 3.1% (P < 0.03). In four animals, REM sleep onset was also accompanied by a transient increase in IBF, ranging from 27 to 106% of NREM baseline. This rise in IBF occurred before the onset of muscle twitches and in parallel with the decrease in MAP. Whereas MAP reached a new, stable baseline, IBF rapidly returned to near-NREM levels. This phenomenon was observed in >75% of REM episodes. Regional blood flows and vascular resistances were similar during QW, NREM sleep, and REM sleep (Table 2).

View this table:
Table 2.

Effect of sleep on hemodynamic parameters

Overall, cardiovascular variables exhibited the greatest variability during REM sleep, but sharp phasic fluctuations were not seen, in contrast with intense phasic rapid eye movements and muscle twitches.

Effect of Arousal on Hemodynamic Variables

With SPA, the following patterned cardiovascular response was commonly observed: renal vasoconstriction and iliac vasodilation, accompanied by tachycardia. This pattern is illustrated for one arousal in Fig.1 and for the group in Figs.2-4. During the 15 s after SPA from NREM sleep, IVR decreased and reached a nadir of −31.2 ± 4.7% from baseline in 9.0 ± 0.4 s, with, consequently, an increase in IBF (+57.5 ± 11.3%). IBF increased before any leg movement, as determined visually. RVR increased and reached a peak of +8.0 ± 4.9% from baseline in 7.2 ± 0.8 s, leading to a decrease in blood flow (−6.5 ± 3.6%). HR increased by 27.3 ± 5.2%, with maximum values achieved in 6.9 ± 1.0 s.

Fig. 1.

Polysomnographic tracing illustrating most common hemodynamic pattern in response to spontaneous arousal from non-rapid-eye-movement (NREM) sleep. Renal vasoconstriction was associated with iliac vasodilation in >50% of NREM spontaneous arousals. bpm, Beats/min.

Fig. 2.

Iliac vascular bed response to spontaneous arousals from NREM sleep, expressed as %baseline. A: iliac blood flow. B: iliac vascular resistance. Each value (mean ± SE) represents mean response for a group of 5 animals.

Fig. 3.

Renal vascular bed response to spontaneous arousals from NREM sleep, expressed as %baseline. A: renal blood flow. B: renal vascular resistance. Each value (mean ± SE) represents mean response for a group of 5 animals.

Fig. 4.

Blood pressure (mean arterial pressure;A) and heart rate (B) response to spontaneous arousals from NREM sleep, expressed as %baseline. Each value (mean ± SE) represents mean response for a group of 5 animals.

After SPA from REM sleep, a pattern of renal vasoconstriction and iliac vasodilation with tachycardia was also observed. In contrast with NREM sleep, however, where changes in MAP after SPA were minimal, MAP increased by 18.1 ± 3.8%, as arterial pressure returned to the wakefulness level after arousal. For HR and IVR, the response in REM sleep was somewhat variable but not statistically different from the response in NREM sleep. Renal vasoconstriction was more intense after arousals from REM sleep but not significantly different from the response in NREM sleep.

AA and TA produced the same pattern of renal vasoconstriction and iliac vasodilation as did SPA. The amplitude of blood flow and HR changes after provoked arousal was not significantly different from the changes after SPA.

Interestingly, not every SPA or provoked arousal was followed by the hemodynamic pattern described above. For the group, approximately one-half of the NREM arousals analyzed (52 ± 6%) were accompanied by a change from baseline in regional vascular resistance of 5% or more (mean change: −32.4 ± 3.8% for IVR and +15.7 ± 4.7% for RVR). Three other regional hemodynamic patterns were observed (Fig. 5). Iliac vasodilation with renal vasodilation occurred in approximately one-third of NREM arousals. The last two patterns (iliac and renal vasoconstriction, iliac vasodilation alone) were seen in 16% of NREM arousals. Similar results were found in REM. As mentioned above, sleep stage (NREM, REM) and the type of arousal (SPA, AA, or TA) did not appear to markedly affect the hemodynamic response that accompanied arousals. When all individual arousals are considered, hemodynamic responses seemed to be of larger amplitude for EEG arousals with marked behavioral response (score ≥4) than for EEG arousals with minimal behavioral response (score ≤3). However, we were unable to perform a formal statistical analysis as two of five pigs did not exhibit arousals with a minimal behavioral response.

Fig. 5.

Distribution of regional vascular resistance patterns after NREM arousals, expressed as %NREM arousals. IVD and RVD, iliac and renal vasodilation, respectively (vascular resistance decrease of ≥5% of baseline); IVC and RVC, iliac and renal vasoconstriction, respectively (vascular resistance increase of ≥5% of baseline).


In this porcine model, we found, first, that sleep did not significantly affect regional renal and iliac hemodynamics, with the exception of a significant decrease in blood pressure in REM sleep. In addition, we found that the cardiovascular response to nonrespiratory arousals was complex and variable. The single most commonly observed pattern consisted of renal vasoconstriction and iliac vasodilation, accompanied by tachycardia. Other patterns included iliac and renal vasodilation, iliac and renal vasoconstriction, or iliac vasodilation alone. Blood pressure response to arousal was variable and small. The local vascular response to arousal was not significantly affected by sleep stage, arousal type, or arousal intensity.

The unique porcine model used in this study was developed by Kirby and co-workers (26, 40) to monitor cardiovascular function during sleep and during airway occlusion. Conscious, unrestrained pigs are used extensively in cardiovascular research. They provide a suitable model for cardiovascular physiology, with anatomic features and control mechanisms during wakefulness close to human characteristics (14, 36). However, the present study demonstrates that some differences exist between humans and juvenile pigs with regard to the effects of sleep on the cardiovascular system. In humans, NREM sleep onset is accompanied by a decrease in HR, leading to a decrease in cardiac output and, consequently, in arterial blood pressure (25). Total peripheral vascular resistance remains unchanged (25). REM sleep is characterized by fluctuations in HR, cardiac output, and arterial blood pressure, with blood pressure values reaching their highest level (19). In contrast, in juvenile pigs, only a slight increase in HR was present in NREM sleep, and arterial blood pressure was maintained. Local vascular resistance was stable in NREM, as is the case in other species (21). With REM onset, blood pressure invariably decreased by ∼15%, and HR accelerated as baroreflexes were stimulated, as our laboratory has previously reported (40). This diminution in blood pressure was accompanied by a slight decrease in renal resistance, likely indicating a global decrease in visceral vascular resistance. A decrease in total peripheral vascular resistance in REM has indeed been reported in the same model (40) and in adults cats (21). Despite these species differences in the effects of sleep on hemodynamics between pigs and humans, insights into the consequences of sudden sleep disruption can be drawn from our study.

Approximately one-half of SPA and induced arousals were accompanied by a specific pattern characterized by iliac vasodilation and renal vasoconstriction. This cardiovascular pattern is similar to the response elicited in mammals during wakefulness by the presentation of a novel and sudden or threatening stimulus (15, 39). Electrical stimulation of the amygdala, the perifornical region of the lateral hypothalamus, or the periaqueductal gray also induces this specific response (6, 10, 15, 32, 35). The cardiovascular reaction to the stimulus is part of a broader behavioral and somatic response, which has been termed the “defense reaction” (or “startle response” if the behavioral component is a startle) (1, 12, 15, 17,32, 33). The behavioral response varies among species but usually consists of an orientation toward the stimulus, which may be preceded by a startle followed by an attack or retreat. The initial cardiovascular component of the defense reaction is characterized by an increase in HR, arterial blood pressure, and cardiac output and by regional vascular changes. Different vascular beds respond differently to the threatening stimulus: vasodilation in the muscular or iliac bed, vasoconstriction in coronary, renal, and mesenteric beds (2, 4, 5, 7,9, 18, 22, 23, 30-33). These cardiovascular changes take place in preparation for the classic “fight or flight” response. This “alerting stage” is quickly followed by hemodynamic changes related to emotion and exercise (39). For some authors, the early hemodynamic response, in particular its renal component, is the main characteristic of the defense reaction (32). In the present model, we found that SPA and provoked nonrespiratory arousals could elicit renal vasoconstriction, iliac vasodilation, and tachycardia, a cardiovascular response similar, if not identical, to that of the alerting stage of the defense reaction. In most species, blood pressure increases transiently after stimulus presentation or electrical stimulation (2,17, 22, 32, 33, 39). This element of the defense reaction was not present in pigs. Interestingly, in the cat, blood pressure does not always increase in the alerting stage of the defense reaction (39). It is conceivable that, in the pig and occasionally in the cat, the decrease in iliac (or muscular) vascular resistance is sufficient to compensate for the increase in renal (or visceral) resistance and the increase in cardiac output mediated by the tachycardia, thereby dampening the blood pressure response. However, we have not yet measured total peripheral vascular resistance to confirm this mechanism. In a previous study using a porcine model with coronary instrumentation, arousals induced by airway occlusions were followed by a small but significant blood pressure increase (+6%) in NREM and a larger increase (+35%) in REM sleep (26). Preliminary results in the present model show a 13% increase in MAP in NREM sleep and a 20% increase in REM sleep after arousals induced by tracheal occlusion (20). Chemostimulation, differences in arousal intensity, or the threatening nature of airway occlusions could contribute to the difference in blood pressure response after nonrespiratory and respiratory arousals.

The cardiovascular response to SPA and nonrespiratory arousals was variable, in terms of magnitude as well as direction. Such variability in the cardiovascular response to external changes has been previously observed in awake pigs (16, 34) and baboons (30). Three methodological aspects of our experimental setup could potentially account for some of the inter- and intra-animal differences. Clearly, circadian variations cannot be totally eliminated, although care was taken to study the animals at approximately the same time of day. Pigs are stress-sensitive animals (34, 36), and a variable level of stress could have repercussions on any cardiovascular response. However, experiments were performed after the animals had had ample time to adjust to laboratory conditions. Finally, after arousal, cardiovascular measurements were obtained over a 15-s period. During that period of time, minor changes in EEG may have been overlooked. Such minor changes in vigilance levels may have influenced the hemodynamic response to arousal and may have contributed to the variability of the response. However, peak changes in cardiovascular variables were attained rapidly, and visual inspection of the EEG trace showed that the animals were awake for at least 10 s after the arousal, even when behavioral changes were minimal. Therefore, we believe that the variability that we observed is not an artifact but, rather, an important feature of the cardiovascular response associated with nonrespiratory arousals in this model. Factors responsible for such hemodynamic variability to arousal remain unknown. In this study, sleep stage, arousal type, and arousal behavioral intensity were the only potential factors that could be examined. None of these factors played a significant role in the response to arousal in this porcine model. Further investigation of the variability of the hemodynamic response to arousal would enhance our understanding of arousal mechanisms.

This study specifically investigated the hemodynamic consequences of SPA. In our model, arousals were classified as “spontaneous” when the investigator failed to perceive any external stimulus within 60 s of the EEG arousal. One cannot eliminate the possibility that some SPA, particularly in NREM sleep, were indeed caused by acoustic stimuli inaudible to the human ear, and this could partly explain the intra- and interanimal variability. Arousals due to undetected respiratory events, such as can be seen in humans, are unlikely to have caused these spontaneous arousals. SPA were observed in animals instrumented with a tracheal catheter, allowing tracheal pressure recordings in this porcine model (20, 26). We did not observe any periodic breathing, spontaneous apneas, or increased negative tracheal pressure leading to arousals in these animals. We believe that the SPA observed in the pigs are comparable to the spontaneous, “normal” arousals recorded in normal adults in a sleep laboratory setting (24). In the present study, we showed that a cardiovascular response can be elicited in the absence of respiratory or nonrespiratory stimulus. Although termed spontaneous, such arousals are likely to result from cortical and/or brain stem activation caused by an internal stimulus. The activation pathways may not be different from those triggered by an external stimulus. However, in the absence of adequate data on arousal mechanisms, the implication of our finding for the neurobiology of the hemodynamic response to sleep disruption remains unclear.

Acoustic and tactile stimuli were also applied during QW, but because the number of events was small, the results were not reported here. When stimuli were associated with a behavioral startle during wakefulness, the patterned cardiovascular response characterized by renal vasoconstriction and iliac vasodilation was elicited. This finding further suggests that the cardiovascular response to arousal parallels the cardiovascular response of the alerting stage of the defense reaction. Repetition of acoustic and, particularly, tactile stimuli during wakefulness and sleep, over the course of a recording session and over the course of the study period, was associated with an attenuated response. Such a phenomenon is characteristic of the startle reaction (12).

In cats and baboons, the renal vasoconstriction characteristic of the defense reaction is usually biphasic (23, 33). The initial decrease is attributed to renal sympathetic activation and the second phase to the release of epinephrine. In our model, only the early phase was present, after SPA and provoked arousals. The renal vasoconstriction in the juvenile pig seems therefore mediated though local neural mechanisms only. Indeed, we have preliminary data showing that, as in other species (23, 33), unilateral denervation abolishes the renal response.

In conclusion, the purpose of this study was to better define the cardiovascular response to SPA and provoked nonrespiratory arousals in pigs. We found that such arousals elicited a patterned cardiovascular response characterized by hindlimb vasodilation and visceral vasoconstriction, although this response was not consistent. This pattern is analogous to the alerting stage of the defense reaction elicited in mammals during wakefulness by presentation of a threatening or sudden stimulus. We speculate that arousals induced by airway occlusion will more consistently elicit this type of hemodynamic response. If confirmed, such findings could have implications for the pathophysiology of the cardiovascular consequences of sleep apnea.


The authors thank the staff of the Animal Research Facility at the West Roxbury Veterans Affairs Medical Center for their help.


  • Address for reprint requests: S. H. Launois, Beth Israel Deaconess Medical Center, East Campus, Dept. of Medicine, 330 Brookline Ave., Boston, MA 02215 (E-mail: slaunois{at}

  • This work was supported by National Heart, Lung, and Blood Institute Grants HL-46829 and HL-48951. At the time of the study, S. H. Launois was a fellow in the Clinical Investigator Training Program, Beth Israel Deaconess Medical Center-Harvard/Massachusetts Institute of Technology Health Sciences and Technology, in collaboration with Pfizer.


View Abstract