Single-session effects of acute intermittent hypoxia on breathing function after human spinal cord injury

Abstract

After spinal cord injury (SCI) respiratory complications are a leading cause of morbidity and mortality. Acute intermittent hypoxia (AIH) triggers spinal respiratory motor plasticity in rodent models, and repetitive AIH may have the potential to restore breathing capacity in those with SCI. As an initial approach to provide proof of principle for such effects, we tested single-session AIH effects on breathing function in adults with chronic SCI. 17 adults (13 males; 34.1±14.5 years old; 13 motor complete SCI; >6 months post injury) completed two randomly ordered sessions, AIH versus sham. AIH consisted of 15, 1-minute episodes (hypoxia: 10.3% O₂; sham: 21% O₂) interspersed with room air breathing (1.5 minutes, 21% oxygen); no attempt was made to regulate arterial CO₂ levels. Blood oxygen saturation (SpO₂), maximal inspiratory and expiratory pressures (MIP; MEP), forced vital capacity (FVC), and mouth occlusion pressure within 0.1s (P_0.1) were assessed. Outcomes were compared using nonparametric Wilcoxon’s tests, or a 2×2 ANOVA. Baseline SpO₂ was 97.2 ± 1.3% and was unchanged during sham experiments. During hypoxic episodes, SpO₂ decreased to 84.7 ± 0.9%, and returned to baseline levels during normoxic intervals. Outcomes were unchanged from baseline post-sham. Greater increases in MIP were evident post AIH vs. sham (median values; +10.8 cmH₂O vs. −2.6 cmH₂O respectively, 95% confidence interval (−18.7) – (−4.3), p=.006) with a moderate Cohen’s effect size (0.68). P_0.1, MEP and FVC did not change post-AIH. A single AIH session increased maximal inspiratory pressure generation, but not other breathing functions in adults with SCI. Reasons may include greater spared innervation to inspiratory versus expiratory muscles or differences in the capacity for AIH-induced plasticity in inspiratory motor neuron pools. Based on our findings, the therapeutic potential of AIH on breathing capacity in people with SCI warrants further investigation.

Introduction

Spinal cord injury (SCI) can cause respiratory muscle weakness and impaired breathing, resulting in recurrent infections, hospitalizations and even death (Cardenas et al., 2004; Krause et al., 2016; National Spinal Cord Injury Statistical Center, 2019). Although diaphragm function is frequently intact, accessory respiratory muscles that contribute to maximal inspiration and expiratory muscles are often weakened. Rehabilitation approaches such as respiratory strength training are effective for increasing breathing function, but deficits persist, and respiratory infection rates remain high for individuals with SCI (Linn et al., 2000; Postma et al., 2009; Raab et al., 2016). Strategies to further promote respiratory motor recovery and improved breathing function are a critical need. One novel strategy to increase respiratory (and non-respiratory) motor function after SCI is low-dose, therapeutic acute intermittent hypoxia (AIH). In humans, AIH protocols typically consist of short episodes breathing low-oxygen air (45–120 sec; 9–10% inspired O₂), alternating with normal inspired oxygen levels (21% O₂) (Gonzalez-Rothi et al., 2015). Although this approach was inspired by years of research on rodent models of AIH-induced respiratory motor plasticity (Fuller et al., 2003; see Gonzalez-Rothi et al., 2015 for review), its first applications in people with chronic SCI concerned its effects on leg strength (Trumbower et al., 2012; Hayes et al., 2014), where its therapeutic efficacy paradoxically appears greater than on resting breathing (Sankari et al., 2015; Tester et al., 2014). However, earlier studies of AIH and breathing in people with SCI focused on resting breathing, and did not consider breathing capacity; further, at least one of those studies had a relatively small sample size (n=8) and utilized a distinct AIH protocol incorporating baseline hypercapnia (high CO₂). Thus, it is unknown if AIH improves breathing capacity in people with SCI.

The purpose of the present study was to test the hypothesis that even a single AIH session can improve breathing capacity in adults with chronic SCI. If verified, this finding would provide important proof-of-principle for subsequent studies with more robust therapeutic AIH protocols, similar to the prior demonstration that a single AIH session increases plantar flexion torque (Trumbower et al., 2012; Lynch et al., 2017; Sandhu et al., 2020) and hand function (Trumbower et al., 2017) which set the stage for subsequent studies of repetitive AIH and effects on standing or walking ability (Hayes et al., 2014; Navarrete-Opazo et al., 2017a,b).

Thus, in the present study, we measured the impact of a single AIH session in people with SCI on 1) maximal voluntary inspiratory and expiratory pressure generation; and 2) forced vital capacity. We also assessed an indicator of resting neuromuscular drive to breathe, mouth occlusion pressure within 0.1s of the onset of inspiration (P_0.1; Whitelaw and Derenne, 1993), presumably reflecting AIH effects on resting neuromuscular drive (respiratory long-term facilitation) with an outcome distinct from tidal volume and breathing frequency (Sankari et al., 2015; Tester et al., 2014). Finally, accessory respiratory muscle activation during maximal voluntary pressure generation was measured via surface electromyograms (EMGs). The impact of AIH on breathing capacity is important since, for example, maximal pressure generation is critical for airway defense (ie. cough) and minimizing respiratory infection risk (Postma et al., 2009; Raab et al., 2016).

Materials and Methods

Participants

Participants were recruited from the community and outpatient centers associated with a large urban rehabilitation health system. Study information was listed on clinicaltrials.gov (NCT03071393). Enrollment criteria included adults ages 18–65 years, who were greater than 6 months post-SCI, C4-T12 levels. Exclusion criteria included self-reported diagnoses of cardiovascular disorders, sleep apnea, other neurologic disorders (i.e., brain injury or stroke), and pregnancy. Study procedures were approved by the University of Florida Institutional Review Board (2016–01680) and were carried out in accordance with the Declaration of Helsinki. All participants provided written informed consent.

Participant Characterization

Participant characteristics such as age, health, and injury history were obtained via selfreport questionnaires. A clinical examination which included upper and lower extremity strength tests and injury classification using the International Standards for the Neurologic Classification of SCI (Kirshblum et al., 2011) was conducted by a licensed physical therapist with expertise in SCI rehabilitation.

Experimental Design

This study was a double-blind, randomized crossover, repeated-measures design. An outline of procedures and an experimental timeline is presented in Figure 1. Participants completed two sessions (AIH and sham sessions), at least 7 days apart. Demographics and the clinical examination were completed during the first session. Breathing measures were obtained before and after AIH and sham (pre- and post-AIH; pre- and post-sham). The delivery of AIH vs. sham was randomized for each participant. Participants and assessors administering breathing assessments were blinded to the AIH or sham intervention assignment.

Figure 1.

Procedures and experimental timeline. After recruitment and consenting, participants were randomized to receive an acute intermittent hypoxia (AIH) or sham on Day 1. Outcomes were collected before and 30 minutes after AIH or sham. At least 7 days later, participants repeated the procedures (Day 2) but received the opposite intervention.

Breathing Assessments

Experimental set-up

Breathing assessments were conducted while the participants reclined with their head and shoulders elevated 45 degrees on a wedge with pillows to support the head and arms as necessary. Breathing was measured using a custom circuit consisting of a flanged rubber mouthpiece attached to a disposable filter and heated neumotachograph (Hans Rudolph, Shawnee, KS, USA) connected to a non-rebreathing valve and amplifier (Hans Rudolph, Shawnee, KS, USA). This circuit enabled continuous breath-by-breath measurement of respired flow, volume and mouth pressure. Breathing assessments adhered to standard guidelines (American Thoracic Society/European Respiratory Society, 2002). Following acclimation to the breathing circuit and procedures, mouth occlusion pressure within 0.1s (P_0.1) during quiet breathing was assessed. Subsequently, tests of maximal inspiratory and expiratory pressure generation and forced vital capacity were performed in a randomized order for each participant.

Resting neuromuscular drive to breathe

Mouth occlusion pressure (P_0.1) is often used as a measure of resting neuromuscular drive at the beginning of a breath (Whitelaw and Derenne, 1993). This measure is relatively independent of the resistance or compliance of the respiratory system, or lung/chest wall sensory feedback associated with a breath. A large bore tube (~1m) was attached to the inspiratory port of the non-rebreathing valve. Following 2 minutes of baseline breathing for familiarization, a study team member out of sight of the participant occluded the open end of the tubing while the participant was expiring. Subsequently, inspiration was briefly occluded (~0.5s). The occlusion was removed once inspiratory pressure exceeded −5.0 cmH₂0. Following this first occlusion, inspiration was randomly occluded every 4–7 breaths until a minimum of 9 occlusions were obtained.

Maximal inspiratory and expiratory pressure generation

Maximal inspiratory mouth pressure (MIP) and maximal expiratory mouth pressure (MEP) are standard measures of inspiratory and expiratory muscle strength, respectively (American Thoracic Society/European Respiratory Society, 2002). To assess MIP, participants exhaled fully while the inspiratory port of the non-rebreathing valve was plugged. Then, participants inhaled forcefully for at least 3 seconds, after which the plug was removed. MEP was assessed in a similar fashion, in that the expiratory port was plugged during a full inhalation, and then participants were asked to forcefully exhale for 3 seconds. A minimum of 4 MIP and MEP maneuvers were completed to ensure 3 tests within 10% or less variability were recorded. Participants rested a minimum of 60 seconds between attempts.

Pulmonary function

Forced vital capacity (FVC) is a standard measure air volume exhaled during a maximal effort following full inspiration (Miller et al., 2005). Using the aforementioned breathing circuit and standard test procedures (Miller et al., 2005), a minimum of 3 trials of FVC were performed to ensure at least 2 trials within 150ml were recorded. Three to five preparatory and recovery breaths were taken before and after each FVC maneuver and participants rested between attempts.

Peak accessory breathing muscle activation

Accessory breathing muscle activity during the maximal pressure generation maneuvers was assessed using surface EMGs recorded bilaterally from the rectus abdominis and the 6^th intercostal space. After shaving excess hair and skin preparation, pairs of Ag/AgCl electrodes (MLA1010, AD Instruments, Colorado Springs, CO, USA) were placed ~2.5 cm apart, centered on the muscle belly (Ovechkin et al., 2010). Rectus abdominis EMGs were recorded at the level of the umbilicus, halfway between the umbilicus and the rib cage. Intercostal EMGs were recorded laterally after manual palpation of the ribs and 6^th intercostal space (Ovechkin et al., 2010). To allow for detection and removal of the electrocardiogram signal from the EMG recordings, a 3-lead ECG signal was recorded.

Signals were continuously recorded at 2,000 Hz (Powerlab 16/35, AD Instruments, Colorado Springs, CO, USA), monitored online using LabChart data acquisition software (Version 8.1, AD Instruments, Colorado Springs, CO, USA) and stored for off-line analysis.

Delivery of AIH or Sham

Participants remained seated, reclined with head and shoulders supported and elevated at 45 degrees for delivery of AIH or sham. The instrumentation and procedures for AIH and sham were identical. A commercially available low-oxygen generator (HYP-123, Hypoxico, New York, NY, USA) and a non-rebreather facemask were used. One minute of low oxygen air or sham was delivered through the facemask, followed by a 90-second bout when the mask was removed to allow room air breathing. This sequence was repeated 15 times. Due to the varied arm and hand impairments in some participants and for consistency, the study staff managed use of the mask to ensure a secure fit and transition between bouts. During AIH, fraction of inspired oxygen (FiO₂) was measured with an oxygen analyzer (Handi+, Maxtec, Salt Lake City, UT, USA) and participants received a 10.3% fraction of inspired oxygen (FiO₂), and 20% FiO₂ was delivered for the sham. Vital signs were monitored for adverse responses. SpO₂ was continuously monitored via a finger pulse oximeter (Infrared Plethysmograph Finger Clip, AD Instruments, Colorado Springs, CO, USA). Blood pressure was monitored via an arm cuff at the end of every other hypoxic or sham interval. Following delivery of AIH or sham, participants rested for 30 minutes and then post-tests were conducted.

Offline Data Processing

Data were processed offline using LabChart software (Version 8.1, AD Instruments, Colorado Springs, CO, USA). Flow, volume, and pressure signals were smoothed using a 0.01-second moving average. To determine P_0.1, a custom script was used to determine inspiration onset and pressure generation 0.1s after onset was quantified. A minimum of eight pressure traces were assessed and averaged for each test session. MIP and MEP were determined by identifying the minimum or maximum value of the pressure trace for each attempt, respectively. FVC was assessed from the volume traces using the Spirometry Module of LabChart. The single best attempt for these maneuvers (MIP, MEP, FVC) was used in the analysis (American Thoracic Society/European Respiratory Society, 2002). This approach was used to capture each individual’s maximal performance or capacity. Automated LabChart scripts were used to detect the fluctuations (peak and trough) of SpO₂ associated with the hypoxic, sham and room air intervals.

EMGs were full-wave rectified and high-pass filtered at 30 Hz. Custom LabChart scripts were used to “blank” the ECG artifacts from the original signals, and a matching length of data from the preceding time window was then copied and inserted into the signal, similar to research devices designed for the same purpose (Chowdhuri et al., 2008). Subsequently, the signal was smoothed using a 6 Hz low-pass filter. Peak EMG amplitudes during the MIP and MEP maneuvers were determined within a window spanning from the inspiration or expiration preceding the MIP or MEP attempt, respectively, to the end of the recovery inspiration or expiration following the MIP or MEP attempt, respectively. This window length was used to ensure that the maximal motor output associated with the maneuvers was captured. Since we sought to determine if motor output was increased post-AIH and individuals with SCI use varied motor strategies, we reasoned that our analysis should broadly include all motor activity associated with the maneuvers. Custom LabChart scripts were used to calculate the average of a 100ms window around the peak amplitude. Peak EMG activity was zeroed by subtracting baseline EMG activity. Baseline EMG activity was based on the mean EMG amplitude calculated from 1s of baseline or quiet activity recorded ~2s before the maneuver. The average of 3 peak amplitudes was calculated for maximal inspiratory and expiratory pressure generation for each participant. Post-intervention EMG amplitudes were normalized to pre-intervention amplitudes, yielding an EMG percent change score for AIH and sham.

Statistics

Breathing outcome data were analyzed to determine if AIH induced a greater improvement from baseline compared to sham. Data were assessed for normality using the Shapiro-Wilk test, and were examined for kurtosis and skewness. FVC outcomes were normally distributed and were assessed using a 2×2 repeated-measures fully-within analysis of variance. MIP, MEP, and P_0.1 outcomes were non-normally distributed. Thus, the effects of AIH versus sham on these outcomes were assessed using Wilcoxon signed-rank tests; separate tests were conducted to assess differences between baseline sham and AIH values and post AIH and sham change scores. To further characterize breathing outcomes following AIH and sham, Cohen’s d effect sizes were calculated using formulas for parametric or non-parametric analyses as appropriate (Ivarsson et al., 2013). Sham and AIH EMG percent change scores were compared using Wilcoxon sign-rank tests. Statistical significance was established at p<.05.

Previous reports have established MIP and MEP correlate with neurologic level of injury (Mateus et al., 2007). Thus, secondary analyses were conducted using Wilcoxon rank-sum tests to determine if baseline MIP, MEP, and post-intervention changes were different for participants with cervical versus thoracic level injuries. Previous reports also have demonstrated an inverse relationship between P_0.1 and MIP in populations with ventilatory weakness (Budweiser et al., 2007; Capdevïla et al., 1995). Spearman’s correlations were used to examine the relationships between MIP and P_0.1 at baseline and between change scores due to AIH or sham.

Statistics were calculated using IBM SPSS Statistics for Windows, version 25 (IBM Corp., Armonk, N.Y., USA). Results are presented as means (parametric analyses) or medians (non-parametric analyses) ± 95% confidence interval (CI). Uncorrected p values are reported (Greenland et al., 2016; Rothman, 1990; Saville, 1990).

Results

Participant Characteristics

Participant characteristics are reported in Table 1. Participants (n=17; 13 males; 34.1±14.5 years old) included individuals with varying injury levels (C4 to T11), severities (American Spinal Injury Association (ASIA) impairment scale (AIS) A, B, C, and D) and durations since injury (7 months to 34 years). Primary outcomes (MIP and MEP) were obtained for all participants. Outcome measures were expanded after the study was initiated and therefore some measures were not conducted for all participants. Participant outcomes are reported in Tables 2–5 and Figures 2–5. All participants completed study procedures as described except in one case where the presence of a colostomy bag prevented EMG recordings from the individual’s left abdominals. No unexpected or serious adverse responses occurred, and procedures were well tolerated.

Table 1.

Participant characteristics.

Participant	Age (years)	Sex	Height (cm)	Weight (kg)	NLI	ISNCSCI	UEMS	LEMS	Time since injury (months)
1	21	M	193	81	C4	D	50	50	38
2	39	M	170	88	T11	A	50	0	21
3	20	M	188	79	T9	A	50	0	9
4	60	M	180	100	C8	B	44	2	15
5	55	M	183	89	C6	A	42	0	104
6	32	F	185	66	T5	B	50	0	162
7	25	M	185	83	T5	B	50	0	19
8	37	M	178	68	T4	A	50	0	25
9	30	M	165	79	T5	A	50	0	31
10	60	F	175	77	C6	A	26	0	8
11	54	M	183	77	C6	B	18	0	410
12	22	M	188	72	C7	B	35	0	7
13	29	M	198	113	T11	C	50	14	8
14	21	M	188	83	C4	B	0	0	70
15	30	F	155	45	T6	B	50	0	108
16	19	M	180	59	C4	C	19	11	21
17	25	F	157	60	T11	C	50	2	16
Average	34.1	-	179.6	77.6	-	-	40.2	4.6	63.1 (5.25 yrs)
Std Dev	14.5	-	11.8	15.9	-	-	15.3	12.4	99.7 (8.3 yrs)

NLI = Neurologic level of injury; ISNCSCI = International Standards for the Neurological Classification of Spinal Cord Injury; UEMS = Upper Extremity Motor Score; LEMS = Lower Extremity Motor Score

Table 2.

Participant outcomes.

	MIP (cmH2O)				MEP (cmH2O)				P0.1 (cmH2O)
Participant	Pre-sham	Pre-AIH	sham dif	AlHdif	Pre-sham	Pre-AIH	sham dif	AIH dif	Pre-sham	Pre-AIH	sham dif	AIH dif
1	77.9	50.2	−2.6	25.5	63.7	71.1	5.4	−11.4	−	−	−	−
2	93.2	99.4	−13.7	10.5	116.5	119.9	−0.1	−3.0	−	−	−	−
3	128.0	132.0	−17.0	−17.0	79.0	74.0	−1.0	10.0	−	−	−	−
4	91.3	81.1	−16.9	16.7	63.8	98.9	12.6	15.4	−	−	−	−
5	62.3	54.2	11.2	10.8	32.3	31.2	12.6	16.5	0.37	0.91	0.10	0.28
6	63.0	49.8	−12.1	11.4	42.2	41.7	−1.5	−5.0	0.76	1.21	0.23	0.69
7	89.1	94.2	15.4	20.9	63.9	84.4	9.8	1.9	−	−	−	−
8	125.9	110.5	−13.0	−10.0	71.7	48.2	−12.5	3.0	1.09	1.01	−0.63	−0.12
9	187.9	158.7	19.7	29.1	125.8	95.6	9.6	21.6	0.30	0.20	0.09	0.29
10	59.9	54.1	0.9	11.9	31.1	28.5	0.3	5.4	1.93	1.65	0.29	−0.28
11	85.8	86.9	−0.5	−10.2	25.8	26.6	4.7	−2.6	0.36	0.48	0.41	0.16
12	96.7	97.5	3.4	8.8	37.6	44.4	14.9	3.1	2.08	1.39	−0.33	0.73
13	122.5	125.2	−13.1	12.5	114.6	174.9	11.1	34.5	0.81	0.77	−0.01	−0.04
14	73.7	74.3	−3.5	−5.8	21.6	27.0	6.3	−3.1	1.21	0.45	−0.09	0.61
15	92.6	112.7	3.8	2.1	82.6	87.5	7.7	−2.8	0.51	0.55	0.11	0.15
16	70.6	74.4	−1.6	10.6	49.1	57.1	−7.5	−3.4	0.55	1.13	0.14	0.17
17	105.4	98.5	−11.7	11.2	100.5	83.8	−23.5	−6.7	1.35	2.50	1.54	0.25
Medians	91.3	94.2	−2.6	10.8	63.8	71.1	5.4	1.9	0.79	0.96	0.11	0.21
95% CI of difference	(−11.5) – 2.7		(−18.7)– (−4.25)		(−13)– 5.9		(−5.4) – 7.7		(−0.49) – 0.31		(−0.48) – 0.29
p value	0.26		0.006		0.35		0.65		0.81		0.43
Cohen's D	n/a		0.68		n/a		0.11		n/a		0.24

MIP = Maximal Inspiratory Pressure; MEP = Maximal Expiratory Pressure; P_0.1 = Mouth Occlusion Pressure; cmH₂O = centimeters of water; dif = difference. “-” indicates outcome measure not completed.

Table 5.

Comparisons between participants with cervical versus thoracic level injuries for maximal inspiratory and expiratory pressure generation (MIP, MEP) baseline values, and changes from baseline due to sham and AIH.

	MIP (cmH2O)
	Sham baseline		AIH baseline		Sham dif		AIH dif
	Cervical	Thoracic	Cervical	Thoracic	Cervical	Thoracic	Cervical	Thoracic
Median	75.8	105.4	74.3	110.5	−1	−12.1	10.7	11.2
95% CI of difference	(−55.3) – (−6.8)		(−60.3) – (13.1)		(−8.5) – 14.6		(−16.3) – 14.3
p value	0.01		0.008		0.48		0.96
	MEP (cmH2O)
	Sham baseline		AIH baseline		Sham dif		AIH dif
	Cervical	Thoracic	Cervical	Thoracic	Cervical	Thoracic	Cervical	Thoracic
Median	34.9	82.6	37.8	84.4	5.6	−0.1	0.25	1.9
95% CI of difference	(−77) – (18.9)		(−67.1) – (−12.7)		(−4.8) – 15.9		(−14.4) – 8.4
p value	0.0006		0.02		0.23		0.88

CI = confidence interval; dif = difference.

Figure 2.

Baseline maximal inspiratory and expiratory pressure generation (MIP, MEP; n=17). Group medians and distributions for sham (white boxes) and AIH (gray boxes) baseline outcomes are shown. cmH₂O = centimeters of water.

Figure 5.

Forced vital capacity outcomes for participants 4–17. Bars representing means and error bars representing 95% confidence intervals are shown for pre and post sham (white bars) and AIH (gray bars).

Delivery of AIH or Sham

Participants completed the two study visits at least 7 days apart (average 18.6 ± 19.4 days). Participants had similar SpO₂ at baseline for sham (97.1 ± 1.4%) and AIH (97.3 ± 1.2%). During AIH, SpO₂ decreased during the hypoxia intervals (84.7 ± 0.9%) and returned to baseline during normoxia (96.7 ± 0.5%). SpO₂ was stable during the sham (96.5 ± 0.3%) and normoxia intervals (97.4 ± 0.5%).

Maximal Inspiratory and Expiratory Pressure Generation (n=17)

Sham and AIH baseline values did not differ for MIP (91.3 vs. 94.2 cmH₂O respectively, 95% CI (−11.5) – 2,7) or MEP (63.8 vs. 71.1 cmH₂O respectively, 95% CI (−13) – 5.9) (Table 2, Figure 2). The AIH-induced change in MIP was significantly greater than change due to sham (+10.8 vs. −2.6 cmH₂O respectively, 95% CI (−18.7) – (−4.3), p = .006) (Table 2, Figure 3). The AIH-induced change in MEP did not differ from change due to sham (+1.9 vs. +5.4 cmH₂O, respectively, 95% CI (−5.4) – 7.7) (Table 2, Figure 3). Consistent with these outcomes, Cohen’s effect size of the MIP change scores (d=0.68) suggests a moderate effect of AIH on MIP (Table 2).

Figure 3.

Changes in maximal inspiratory and expiratory pressures. Group medians and distributions for sham (white boxes) and AIH (gray boxes) changes from baseline pressures are shown. * = significant change (p<.01). cmH₂O = centimeters of water.

Peak Accessory Breathing Muscle Activation (n=12)

No significant differences in AIH- or sham-induced EMG amplitude change scores were evident in the left or right intercostals or abdominal muscles (Table 3).

Table 3.

EMG results.

	MIP Changes (%)			MEP Changes (%)
	sham		AIH	sham		AIH
L Int 95% CI	−5		+11	−8.5		+1
	(−65) – 29			(−22) – 42
R Int 95% CI	4		1	+13.5		−8.5
	(−49) – 44			(−2) – 47
L Ab 95% CI	−12		−7	−8		+17
	(−67) – 48			(−52) – 116
R Ab 95% CI	−9.5		+44.5	+9		−8
	(−131) – 4			(−296) – 51

Percent changes in peak amplitude from baseline are presented as medians with 95% confidence intervals. Data are from participants 6–17 (n=12). L = left; R = Right; Int = Intercostals; Ab = Abdominals

Resting Neuromuscular Drive to Breathe (n=12)

Sham versus AIH baseline values for P_0.1 did not differ (0.79 vs. 0.96 cmH₂O respectively, 95% CI (−0.49) – 0.31) (Table 2, Figure 4A). The AIH-induced change in P_0.1 did not significantly differ from sham (0.21 vs. 0.11 cmH₂O, respectively, 95% CI (−0.48) – 0.29) (Table 2, Figure 4B).

Figure 4.

Mouth occlusion pressure (P_0.1) outcomes for participants 5, 6, and 8–17. Group medians and distributions for baseline (A) and changes from baseline (B) pressures for sham (white boxes) and AIH (gray boxes) are shown. cmH₂O = centimeters of water.

Pulmonary Function (n=14)

No differences were found for FVC (F(1,13) = .027, p=.872) for all time points and conditions (Table 4, Figure 5).

Table 4.

Individual results for forced vital capacity (n=14).

Participant	Pre-sham	FVC (L)		Post-AIH
		Post-sham	Pre-AIH
1	-	-	-	-
2	-	-	-	-
3	-	-	-	-
4	3.11	2.97	3.07	2.84
5	1.98	1.83	2.28	2.32
6	2.61	2.80	2.71	2.66
7	4.36	4.45	4.80	4.75
8	4.00	3.47	3.77	3.63
9	4.15	4.16	4.02	4.39
10	1.53	1.72	1.65	1.66
11	2.05	2.15	1.82	1.87
12	2.80	3.16	2.79	2.86
13	5.91	5.86	6.01	5.87
14	2.27	2.15	2.71	2.34
15	3.21	3.25	3.16	3.27
16	2.90	3.04	2.98	3.09
17	2.75	2.53	2.50	2.51
Average	3.11	3.11	3.16	3.15
Std dev	1.16	1.13	1.17	1.17
95% CI	0.61	0.59	0.61	0.61

Secondary Analyses

Effects of injury level on baseline MIP and MEP values and responses to AIH versus sham are summarized in Table 5. Participants with cervical injuries (n=8) had significantly lower baseline MIP and MEP values than participants with thoracic injuries (n=9). There were no differences between responses to AIH and sham for MIP and MEP in individuals with cervical versus thoracic injuries.

There was no correlation between MIP and P_0.1 values at baseline (sham, r=−.08, p>.05; AIH, r= −.39, p>.05). In general, participants with lower MIP outcomes had higher P_0.1 outcomes. There was no relationship between change in MIP and change in P_0.1 after either intervention (sham, r=.06, p>.05; AIH, r=.04, p>.05).

Discussion

Our results indicate that a single AIH session increases maximal inspiratory pressure generation in adults with SCI. This suggests the ability to generate a forceful, voluntary inspiratory maneuver was increased in people with heterogeneous injury characteristics and varied breathing impairments. AIH did not consistently affect resting neuromuscular drive to breathe (P_0.1), maximal expiratory pressure generation or pulmonary function (FVC). There were no detectable changes in peak intercostal or abdominal activation during maximal pressure generation tests. Potential reasons for these differential AIH responses and implications are discussed.

Differential effects of AIH on maximal inspiratory versus other breathing functions

In this heterogeneous group of individuals with SCI, all who live in the community and breathe independently, AIH improved maximal inspiratory, but not expiratory function. One possible explanation for this differential outcome may be greater sparing of descending axonal pathways to the diaphragm and other accessory inspiratory muscles in these participants. Expiratory motor neurons exit the spinal cord caudal to the phrenic motor nucleus, and therefore expiratory function is expected to be more impaired in this heterogeneous population.

Residual descending pathways (based on injury level, segments serving inspiratory vs. expiratory function) are a substrate upon which AIH can elicit serotonin release and, potentially, amplify motor output (Golder and Mitchell, 2005). However, since AIH (or carotid sinus nerve stimulation) induced long-term facilitation is more prominent in inspiratory versus expiratory nerves (Mitchell et al., 2001; Navarrete-Opazo et al., 2014), mechanisms necessary for AIH-induced respiratory motor plasticity may be lacking in expiratory (versus inspiratory) motor neuron pools.

Based on injury levels and clinical examination, all participants, even those diagnosed with motor complete SCI, had intact innervation to at least some accessory inspiratory muscles. Sternocleidomastoid and upper trapezius muscles often contribute to inspiratory pressure generation in individuals after SCI (Aslan et al., 2013; De Troyer et al., 1986; Ovechkin et al., 2010), and may have exhibited AIH-induced facilitation; however, EMG activity was not recorded in these muscles. On the other hand, participants likely had differential innervation of the intercostal muscles and variable muscle activation patterns during MIP generation. These individual differences would affect the potential for AIH-induced plasticity in intercostal muscle activation (EMGs), which was not consistently detected in this study and may explain why level of injury was not related to AIH-induced changes in MIP.

Despite greater AIH-induced increases in P_0.1 (vs. sham) in eight of twelve participants, P_0.1 changes did not statistically differ between conditions. While both P_0.1 and MIP rely on inspiratory neuromuscular pathways, MIP relies on maximal voluntary motor recruitment, whereas P_0.1 quantifies the automatic response to an unexpected perturbation during quiet breathing, i.e. within 0.1 seconds. Thus, MIP and P_0.1 evaluate different aspects of respiratory function which likely explains why baseline outcomes and post-intervention outcomes were not correlated.

A potential reason why P_0.1 changes were not detected in all participants is the absence of elevated arterial carbon dioxide (P_CO2), i.e., hypercapnia. Our study protocol manipulated O₂, but CO₂ was not manipulated or controlled. AIH-induced increases in ventilation in humans with SCI, which reflect increased drive to breathe has only been shown under hypercapnic conditions (Tester et al., 2014; Sankari et al., 2015; Mateika et al., 2018). Hypercapnia alone elevates resting ventilation, and thus may augment or alter AIH-induced effects on ventilation. Whether AIH would increase P_0.1 under hypercapnic conditions requires further experimentation.

A single AIH session did not alter expiratory function

In contrast to MIP outcomes, a single AIH session did not change expiratory volume or pressure generation (i.e., FVC and MEP). As expected, baseline MEP generation was lower in participants with cervical versus thoracic injuries. However, responses to AIH and sham did not differ between those with cervical versus thoracic injuries. One explanation of these findings is greater variability of motor strategies (i.e., pattern of muscle activation) used by people with SCI to perform maximal expiratory pressure generation versus strategies used during MIP maneuvers (Ovechkin et al., 2010). Further, motor strategies are more variable in those with severe motor impairments (Ovechkin et al., 2010). Pursuant to this point, the majority of participants in this study had motor complete injuries (AIS A or B). Thus, motor strategies used during maximal expiratory maneuvers were likely variable, causing inconsistent outcomes and difficulty detecting expiratory changes.

Another consideration pertaining to the expiratory outcomes is altered breathing mechanics, such as reduced lung and rib cage compliance, which is common after SCI (Schilero et al., 2009). In this study, most participants had chronic injuries (13 of 17 participants had injuries >1 year) and reduced FVCs suggesting that many participants had altered breathing mechanics. FVC and altered breathing mechanics which result from long-term changes in the lung tissue and stiffness in the ribcage joints (Schilero et al., 2009) are likely to be unchanged by a single intervention session.

Limitations and Future Directions

A limitation of this study is that participants’ injury characteristics and demographics were diverse, limiting understanding of factors such as age or sex, which may affect AIH-induced plasticity (McGuire and Ling, 2005; Zabka et al., 2005) and association with other factors such as sleep apnea that vary by SCI injury level and severity (Sankari et al., 2015). Participant exclusion for sleep apnea was based on self-report rather than an overnight examination. For this proof-of-concept study, this approach was used to minimize participant burden. However, the presence of sleep apnea is thought to be grossly under-reported in people with cervical SCI (Graco et al., 2018), and sleep apnea may have influenced study outcomes (Vivodtzev et al., 2020). While participant heterogeneity constrains interpretation of study outcomes, this also suggests that the study findings of AIH-induced effects on MIP are robust and warrant further consideration.

Another possible limitation is that medication use was not constrained or closely tracked. However, participants were asked to maintain a consistent routine and medication use on both testing days. AIH effects and impact of common medications prescribed after SCI, such as anti-depressants, anti-spasticity medication or medications for neuropathic pain, are an important consideration (Lynch et al., 2017; Sandhu et al., 2019). Further, protocol features such as duration of hypoxic and normoxic intervals or level of hypoxia have not been optimized, nor did we use background hypercapnia, which is reported to amplify ventilatory long-term facilitation in people (Mateika et al., 2018). In this study, outcomes were assessed 30-minutes post AIH (or sham), in part to reduce participant burden (vs. multiple measures at several time points). It is unknown if effects would persist (or increase) at later times post-AIH. Finally, the therapeutic potential of multiple sessions of AIH also cannot be inferred from these results, as the functional impact of AIH is enhanced with repeated sessions (i.e., preconditioning; Fields and Mitchell, 2015).

With respect to non-respiratory motor behaviors, AIH effects are enhanced when paired with task specific training (Hayes et al., 2014; Prosser-Loose et al., 2015; Welch et al., 2020). Similar synergistic relationships between AIH and task specific training have not been established for breathing. Thus, it remains possible that combinatorial treatments with AIH plus breathing exercises such as respiratory strength training, isocapnic voluntary hyperventilation or hypercapnic hyperpnea may synergistically augment the impact of AIH on inspiratory capacity and may reveal plasticity in expiratory activity. This possibility awaits direct experimental verification.

Mechanisms of AIH-induced respiratory plasticity

Although the present study provides no specific insights concerning underlying mechanisms of respiratory motor plasticity, abundant evidence is available from studies of AIH-induced phrenic motor plasticity in rodent models (Mitchell et al., 2001; Feldman et al., 2003; Devinney et al., 2013; Dale-Nagle et al., 2010, 2014; Gonzalez-Rothi et al., 2015). In brief, phrenic long-term facilitation following a single AIH presentation requires activation of raphe, serotonergic neurons, serotonin release and type 2 serotonin receptor activation on or near phrenic motor neurons (Baker-Herman and Mitchell, 2002; MacFarlane et al., 2011; Tadjalli and Mitchell,2019), new synthesis of brain derived neurotrophic factor (Baker-Herman et al., 2004) and activation of its high affinity receptor (TrkB) within phrenic motor neurons (Dale et al., 2017). A competing mechanism associated with AIH either constrains this serotonergic mechanism of phrenic/diaphragm motor plasticity (Hoffman et al., 2010; Hoffman and Mitchell, 2013; Navarrete-Opazo et al., 2014), but can replace it when hypoxemia is severe during the AIH protocol (Nichols et al., 2012). The mechanism of respiratory long-term facilitation most often studied in humans is presumed to be the serotonin-dependent mechanism, but this hypothesis remains to be rigorously tested.

Conclusion

In a heterogeneous sample of adults with SCI, a single AIH session augments maximal inspiratory pressure generation (MIP). Since maximal inspiratory pressure generation is a strong predictor of risk for pneumonia after SCI (Raab et al., 2016), AIH may be a potential non-invasive therapeutic strategy to improve aspects of breathing that are relevant to the health of those living with SCI.