Impact of diabetes mellitus on the respiratory function of amyotrophic lateral sclerosis patients

Abstract

Background and purpose

Respiratory insufficiency and its complications are the main cause of death in amyotrophic lateral sclerosis (ALS). The impact of diabetes mellitus (DM) on respiratory function of ALS patients is uncertain.

Methods

A retrospective cohort study was carried out. From the 1710 patients with motor neuron disease followed in our unit, ALS and progressive muscular atrophy patients were included. We recorded demographic characteristics, functional ALS rating scale (Amyotrophic Lateral Sclerosis Functional Rating Scale–Revised [ALSFRS‐R]) and its subscores at first visit, respiratory function tests, arterial blood gases, phrenic nerve amplitude (PhrenAmpl), and mean nocturnal oxygen saturation (SpO₂mean). We excluded patients with other relevant diseases. Two subgroups were analysed: DIAB (patients with DM) and noDIAB (patients without DM). Independent t‐test, χ ², or Fisher exact test was applied. Binomial logistic regression analyses assessed DM effects. Kaplan–Meier analysis assessed survival. p < 0.05 was considered significant.

Results

We included 1639 patients (922 men, mean onset age = 62.5 ± 12.6 years, mean disease duration = 18.1 ± 22.0 months). Mean survival was 43.3 ± 40.7 months. More men had DM (p = 0.021). Disease duration was similar between groups (p = 0.063). Time to noninvasive ventilation (NIV) was shorter in DIAB (p = 0.004); total survival was similar. No differences were seen for ALSFRS‐R or its decay rate. At entry, DIAB patients were older (p < 0.001), with lower forced vital capacity (p = 0.001), arterial oxygen pressure (p = 0.01), PhrenAmpl (p < 0.001), and SpO₂mean (p = 0.014).

Conclusions

ALS patients with DM had increased risk of respiratory impairment and should be closely monitored. Early NIV allowed for similar survival rate between groups.

INTRODUCTION

Respiratory insufficiency (RI) and its complications are the main cause of death in amyotrophic lateral sclerosis (ALS). Respiratory symptoms emerge when the respiratory muscles are weakened by involvement of the neuronal pathways innervating them. Inspiratory involvement is reported as dyspnoea for progressively milder physical activities, eventually at rest, and orthopnoea [1]. Expiratory muscle involvement causes impaired cough efficacy leading to decreased airway clearance and greater risk of pulmonary infection [2].

Diabetes mellitus (DM) is a chronic metabolic disorder that involves many organs and systems. In the nervous system, focal and multifocal neuropathies and generalized neuropathies are common complications of DM [3, 4]. Phrenic neuropathy has been described in patients with DM [5, 6, 7]. Several studies have addressed the association of DM with ALS. The risk of ALS seems to be higher in subjects with DM diagnosed before the age of 40 years; otherwise, DM type 2 seems possibly to have a protective role on ALS onset [8, 9, 10, 11, 12, 13]. In other studies, DM does not impact on the severity, progression, and survival of ALS [14], not even showing any epidemiological association [15]. Some reports describe a protective effect of obesity, higher body mass index (BMI), hypercholesterolaemia, and hyperlipidaemia on ALS [16, 17, 18, 19, 20, 21, 22, 23], and these factors have also been associated with longer survival. This metabolic profile is commonly associated with DM. Two thirds of ALS patients develop hypermetabolism, and fast weight loss from disease onset to the first visit is associated with poor prognosis [18, 24]. Furthermore, hyperlipidaemia, with a high low‐density to high‐density lipoprotein rate, correlates with better prognosis, probably by supporting these higher energy requirements [17].

The impact of DM on respiratory function in ALS is unknow. Because phrenic nerve dysfunction is associated with hypoventilation in ALS [25], we wondered whether DM might be associated with more rapid respiratory impairment in ALS.

METHODS

Subjects

A retrospective cohort study was carried out. From the 1710 patients with motor neuron disease followed in our unit (1995–2022) and registered in our database, we included those with ALS and progressive muscular atrophy, using the Gold Coast diagnostic criteria [26]. Patients with respiratory diseases, anaemia, heart failure, cancer, and marked cognitive involvement were excluded. A retrospective cohort study of patients followed prospectively was carried out. The whole population was divided into two groups, depending on the presence or absence of DM as a concomitant diagnosis: group 1 (DIAB) including patients with DM and subgroup 2 (noDIAB) including patients without DM. Patients classified as DM had fasting blood sugar levels > 126 mg/dL (7 mmol/L) on two separate tests, and had been medically treated for this condition for >6 months at the time of ALS onset (recorded in the database with the codes 0 and 1: 0, no DM; 1, DM).

Clinical evaluation

We recorded the demographic characteristics of the whole population at baseline, and of the two subgroups, as well as the following functional parameters and clinical tests, which were all carried out within a 1‐month time frame.

Functional parameters

The Amyotrophic Lateral Sclerosis Functional Rating Scale–Revised (ALSFRS‐R) and its subscores at baseline visit were recorded, namely the bulbar, upper limb, and lower limb scores as well as the respiratory subscore [27]. The total ALSFRS‐R score is rated from 0 to 48 (min–max functioning score), and each of the subscores is rated from 0 to 12 (min–max functioning score). The decay of ALSFRS‐R per month was also evaluated and determined as (48 − ALSFRS‐R at baseline)/disease duration at baseline.

Respiratory function tests

Respiratory function tests were performed with the patients in the sitting position, using a spirometer and whole‐body plethysmography, according to the American Thoracic Society recommendations [28]. We recorded the best of three satisfactory and consistent expiratory manoeuvres, performed forcefully (forced vital capacity [FVC]) and slowly (slow vital capacity [SVC]) and obtained after a maximal inspiratory effort [28]. In addition, the maximal inspiratory (MIP) and expiratory (MEP) muscle pressures were assessed and recorded as the best result of three consistent measurements at the mouth, generated against an occluded airway [29]. The inspiratory pressure at 100 ms after an occluded inspiratory effort (P0.1) was also assessed [30]. The maximal values for FVC, SVC, MIP, MEP, and P0.1 were compared with the predicted values from a control group matched for age, gender, and height. For each variable, the resulting percentage was used in the analysis.

Blood gas analyses

Blood gas analysis was performed before the other respiratory tests, while the patients, in the sitting position, had been breathing room air for at least 30 min. Arterial oxygen tension (pO₂), arterial carbon dioxide tension (pCO₂), pH, and bicarbonate ion (HCO₃ ^‐) were measured with an automated analyser (ABL 500; Radiometer, Copenhagen, Denmark) [31].

Nocturnal percutaneous pulse oximetry

Mean oxygen saturation (SpO₂mean) was recorded continuously during sleep by fingertip infrared pulse oximeter. A minimum of 6 h of nocturnal recording was required [32].

Neurophysiology

Diaphragmatic motor responses were elicited by percutaneous bipolar electrical phrenic nerve stimulation in the neck and recorded with surface electrodes (filter setting 20 Hz–10 kHz), with the active electrode positioned at the homolateral costosternal angle and reference electrode positioned 16 cm distant at the costal margin. A minimum of five consistent motor responses were recorded, and the one with the highest peak‐to‐peak amplitude from each side (right and left) was selected. The mean maximal amplitude of the diaphragmatic motor response was used in the statistical analyses (phrenic nerve amplitude [PhrenAmpl]), or the categorical normal versus abnormal PhrenAmpl (abnormal if <0.4 mV) [33].

Statistical analyses

Two investigators (S.P. and M.G.) have continuously updated the database (from 2006 to date). Both regularly check the full database for irregularities, including incongruencies in data insertion. A final check was done by S.P., who went through all variables, including doing demographic analyses to detect deficiencies in data insertion.

Independent t‐test for continuous data and χ ² or Fisher exact test for categorical data was used for analysing the differences between the two subgroups regarding the studied parameters. A binomial logistic regression analysis, with backward method, was performed to determine the independent predictors of having or not having DM. Death or censor date (31 May 2022) was considered as the event. Differences between survival from first symptoms to noninvasive ventilation (NIV), death or censor date, as well as total survival from first symptoms to death or censor date, were analysed using the Kaplan–Meier curves with log‐rank (Mantel–Cox) analyses. Sensitivity, specificity, and positive (PPV) and negative (NPV) predictive values of having or not having DM for respiratory involvement (RI) as determined by a normal or abnormal PhrenAmpl (abnormal if <0.4 mV) were determined, after performing crosstabs, with χ ² analyses.

Analyses were performed in SPSS v24.0 (IBM, Armonk, NY), and p < 0.05 was considered as meaningful.

Ethical committee

All procedures performed in the study involving human participants were carried out in accordance with the ethical standards of the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. This study was approved by the local ethics committee (ref 162/21).

RESULTS

The entry criteria were met by 1639 patients, of whom 922 were men, with a mean age at disease onset of 62.5 ± 12.7 years, and a mean disease duration of 18.1 ± 22.0 months at evaluation. Onset was spinal in 1081, bulbar in 448, respiratory in 65, axial in 29, and generalized in 10 (not identified in six). Total survival was 43.3 ± 40.7 months. Type 2 DM was present in 167 (10.2%) patients, who were therefore included in DIAB. In those, 15% (25 patients) had mild to moderate polyneuropathy, characterized by absent or small sensory nerve action potentials in lower limbs, in general associated with moderately reduced nerve conduction velocities and delayed F‐waves in the legs. No patient had severe polyneuropathy (defined as absent motor responses in lower limbs). No patient had type 1 DM. Of the 1639 patients, a total of 1250 patients were assessed with PhrenAmpl and 485 (38.8%) had an abnormal PhrenAmpl at baseline. For demographic characteristics of the patients included, please refer to Table 1.

TABLE 1.

Demographic characteristics of the total population and G1 and G2 subgroups as well as functional and other respiratory parameters.

Characteristic	Total population, n = 1639, mean ± SD, min–max	G2: no DM, n = 1472, mean ± SD, min–max	G1: with DM, n = 167, mean ± SD, min–max	p
Male gender, n	922	814	108	0.021*
BMI at 1st visit, kg/m2	24.8 ± 3.9	24.7 ± 3.8	25.8 ± 4.0	0.002**
	13.4 to 39.7	13.4 to 39.7	18.37 to 38.8
Disease duration, months	18.1 ± 22.0	17.7 ± 21.4	21.1 ± 26.5	0.063
	0.7 to 282.0	0.7 to 282.0	1.02 to 215.0
Onset age, years	62.5 ± 12.7	62.1 ± 13.0	66.5 ± 8.5	<0.001**
	9 to 90	9 to 90	45 to 85
Time to NIV, months	9.2 ± 16.5	9.8 ± 17.0	4.7 ± 10.2	0.004**
	−18 to 187	−18 to 187	−15.1 to 53.0
Survival with NIV, months	20.0 ± 24.9	20.2 ± 25.4	18.0 ± 21.0	0.418
	−60.3 to 215	−60.3 to 215	−0.69 to 145.2
Total survival, months	43.3 ± 40.7	43.6 ± 41.4	40.7 ± 34.2	0.388
	2.1 to 412.94	2.1 to 412.94	7.49 to 232.64
ALSFRS	30.4 ± 6.8	30.5 ± 6.9	29.7 ± 6.2	0.167
	2 to 40	2 to 40	8 to 40
ALSFRS‐R	38.1 ± 7.2	38.2 ± 7.2	37.2 ± 6.6	0.095
	9 to 48	9 to 48	14 to 48
ALSFRS‐R decay rate	0.963 ± 1.14	0.959 ± 1.16	0.995 ± 0.96	0.705
	0 to 16.5	0 to 16.5	0 to 5.93
ALSFRSb	10.0 ± 2.5	10.0 ± 2.6	10.2 ± 2.5	0.313
	0 to 12	0 to 12	2 to 12
ALSFRSul	8.6 ± 3.2	8.7 ± 3.2	8.4 ± 3.1	0.303
	0 to 12	0 to 12	0 to 12
ALSFRSll	8.3 ± 3.2	8.3 ± 3.2	7.7 ± 3.2	0.014*
	0 to 12	0 to 12	0 to 12
RofALSFRS‐R	11.1 ± 1.6	11.1 ± 1.5	10.9 ± 1.9	0.072
	3 to 12	3 to 12	3 to 12
SVC, %predict	82.6 ± 23.7	83.5 ± 24.0	74.6 ± 20.1	0.001**
	18.0 to 160.0	18.0 to 160.0	24.0 to 120.6
FVC, %predict	83.8 ± 24.2	84.6 ± 24.4	76.5 ± 20.8	0.001**
	15.0 to 166.0	15.0 to 166.0	18.0 to 125.0
MIP, %predict	54.1 ± 29.2	54.7 ± 29.5	48.9 ± 26.1	0.084
	4.6 to 175.0	4.6 to 175.0	12.93 to 132.0
MEP, %predict	63.8 ± 29.4	64.2 ± 29.0	59.8 ± 33.7	0.202
	5.1 to 186.6	5.1 to 175	5.6 to 186.6
P0.1, % predict	98.8 ± 47.3	98.2 ± 47.4	105.8 ± 45.7	0.275
	0.3 to 370.4	0.3 to 370.4	45.0 to 272.1
pO2	85.0 ± 9.8	85.3 ± 9.7	81.6 ± 10.5	0.01*
	57.0 to 113.2	57.0 to 113.2	60.0 to 111.7
pCO2	40.0 ± 5.3	40.0 ± 5.3	40.2 ± 5.4	0.859
	28.1 to 99.6	28.1 to 99.6	31.5 to 57.0
pH	7.43 ± 0.02	7.4 ± 0	7.4 ± 0.03	0.575
	7.34 to 7.52	7.3 to 7.5	7.39 to 7.48
PhrenAmpl, mV	0.49 ± 0.29	0.51 ± 0.29	0.38 ± 0.27	<0.001**
	0 to 2.2	0 to 2.2	0 to 1.3
SpO2mean, %	94.1 ± 2.3	94.2 ± 2.1	93.3 ± 4.0	0.014*
	71.3 to 98.1	75.2 to 98.1	71.3 to 98.0

Note: p < 0.05 was considered significant.

Abbreviations: %predict, percent predicted; ALSFRS, Amyotrophic Lateral Sclerosis Functional Rating Scale; ALSFRSb, ALSFRS bulbar score; ALSFRSll, ALSFRS lower limb score; ALSFRS‐R, ALSFRS–Revised; ALSFRSul, ALSFRS upper limb score; BMI, body mass index; DM, diabetes mellitus; FVC, forced vital capacity; MEP, maximal static expiratory pressure; MIP, maximal static inspiratory pressure; NIV, noninvasive ventilation; P0.1, pressure at 100 ms of an inspiratory effort against a closed airway; pCO₂, arterial carbon dioxide tension; PhrenAmpl, mean peak‐to‐peak amplitude of phrenic nerve motor responses in right‐ and left‐sided recordings; pO₂, arterial oxygen tension; RofALSFRS‐R, ALSFRS‐R respiratory subscore; SpO₂mean, mean nocturnal pulsed oxygen saturation; SVC, slow vital capacity.

*p < 0.05, **p < 0.01.

DIAB patients were older (66.5 ± 8.5 vs. 62.1 ± 13.0 years, p < 0.001), predominantly men (p = 0.021), had higher BMI at baseline visit (25.8 ± 4.0 vs. 24.7 ± 3.8 years, p = 0.002); they had respiratory onset more frequently (p = 0.001), more men were affected (p = 0.021). Spinal versus bulbar versus axial onset region was not different between the two groups (p > 0.05).

Disease duration at evaluation and total survival were also similar (p > 0.05), but patients in DIAB were adapted sooner to NIV (4.7 ± 10.2 vs. 9.8 ± 17.0 months, p = 0.004). No differences were seen between DIAB and noDIAB regarding functional scores, including the decay of ALSFRS‐R and the respiratory subscore of the ALSFRS‐R at baseline (p > 0.05), but lower limbs subscore was significantly lower in patients with DM (p = 0.014). DIAB patients had lower FVC (76.5 ± 20.8% vs. 84.6 ± 24.4%, p = 0.001), lower pO₂ (81.6 ± 10.5 mmHg vs. 85.3 ± 9.7 mmHg, p = 0.01), lower SpO₂mean (93.3 ± 4.0 vs. 94.2 ± 2.1, p = 0.014), and lower PhrenAmpl (0.38 ± 0.27 mV vs. 0.51 ± 0.29 mV, p < 0.001) than noDIAB patients. No significant correlations were found between the investigated respiratory parameters and BMI in the DIAB group (p > 0.05).

In a binomial logistic regression analysis, where all significant variables were included, only PhrenAmpl (introduced as a categorial variable, normal/abnormal) was an independent predictor of having DM (exponential of B = 0.113, 95% confidence interval = 0.037–0.343, p < 0.001), indicating that for those patients with abnormal PhrenAmpl the odds of having DM increased by 88.7%.

Kaplan–Meier curves with log‐rank analyses, showed significant differences between groups regarding time to NIV (χ ² = 9.042, p = 0.003) but not regarding survival with NIV and total survival (χ ² = 1.256, p = 0.263 and χ ² = 2.74, p = 0.098, respectively) as determined by log‐rank Mantel–Cox analyses (Figure 1).

FIGURE 1.

Survival curves between the two groups of patients (with and without diabetes mellitus [DM]). (a) Time from first symptoms to noninvasive ventilation (NIV) adaptation. (b) Time from first symptoms to death or censor date. No differences were noted (p > 0.05). Cum, cumulative; DIAB, patients with DM; noDIAB, patients without DM.

Regarding having RI or not as determined by abnormal or normal PhrenAmpl values, respectively, the χ ² analyses revealed p < 0.001, with the sensitivity and specificity of having or not having DM for the presence of RI of 15.26% and 93.2%, respectively; PPV and NPV were 58.73% and 63.43%, respectively. Normal (=12) or abnormal values (<12) of the respiratory subscore of the ALSFRS‐R did not predict having or not having DM (p = 0.842; Figure 2).

FIGURE 2.

Bar chart representing, in patients with and without diabetes mellitus (DM), (a) the number of patients with normal (≥0.4 mV) or abnormal (<0.4 mV) phrenic nerve studies and (b) the number of patients with normal (=12) or abnormal (<12) respiratory subscore of the Amyotrophic Lateral Sclerosis Functional Rating Scale–Revised (RofALSFRS‐R). G1, no DM; G2, with DM; PhrenAmpl, phrenic nerve amplitude.

DISCUSSION

The impact of DM on respiratory function of ALS patients has not previously been studied. In our analysis, ALS patients with DM (DIAB) presented a distinct phenotype as compared to those without DM (noDIAB). DIAB included significantly more men, older patients with higher BMI, and importantly, patients with more severe respiratory involvement at baseline. This more severe involvement was supported by the more frequent respiratory onset phenotype in DIAB and lower values in the respiratory function tests, which were significant for FVC (p = 0.001) and mean PhrenAmpl (p < 0.001), associated with diurnal (pO₂, p = 0.01) and nocturnal (SpO₂mean, p = 0.014) hypoxemia. A trend for a compensatory higher central respiratory activity, as investigated by P0.1 values, was recorded. This compensatory mechanism has been reported before in ALS [34] and could explain, to some extent, the nonsignificant differences in the respiratory subscore of the ALSFRS‐R. Patients in DIAB were adapted 5 months earlier to NIV (p = 0.004), but no differences in survival were found between DIAB and noDIAB, supporting the role of NIV as an effective intervention for respiratory involvement in ALS, including patients whose ALS was associated with DM.

In a binomial logistic regression analysis, only PhrenAmpl was an independent predictor of being included in the group with DM (p < 0.001), indicating that for those patients with abnormal PhrenAmpl the odds of having DM increased by 88.7%. PhrenAmpl recorded at baseline has been shown previously to be predictive of hypoventilation [25] and survival [33] in our ALS patients. Our current study extends the relevance of performing phrenic nerve studies as part of the respiratory tests at the time of the first visit.

In our view, there are two main hypotheses to explain the negative effect of DM on respiratory function in ALS patients. The first is that patients with DM have a greater chance of phrenic nerve dysfunction due to the demise of spinal motor neuronal pools. The second is that the metabolic dysfunction related to DM causes a higher risk of axonal degeneration of the phrenic nerve. Patients with DM have an increased risk of cognitive dysfunction, as a consequence of neuronal death and synaptic abnormalities, probably derived from microcirculatory changes and diabetes‐related oxidative stress, which has been well studied in the hippocampus [35]. In animal models, DM reduces the number of motor neurons [36], and in patients with DM magnetic resonance imaging studies have revealed atrophy of motor‐related areas, not related to cardiovascular risk factors [37, 38, 39]. In addition, a reduction in the volume of the corticospinal tract has been reported on imaging in patients with DM [40], possibly related to increased central conduction time on transcranial magnetic stimulation in these patients [41]. On the other hand, DM commonly affects the axonal function, in particular causing axonal swelling related to oxidative stress, and increased polyol pathway and reduced Na⁺/K⁺‐ATPase activity [43, 44]. These changes compromise axonal transport, and abnormal retrograde transport can damage spinal motor neurons by causing abnormality in reception of essential trophic factors, as reported in an animal model of diabetes [44, 45, 46]. Chronic axonal dysfunction in DM causes structural changes in the axons, in particular degeneration of the nodal myelin and loss of axoglial junctions, with repercussions on paranodal function [42, 43, 47]. This could support a local metabolic dysfunction, as opposed to the first explanation, a dying forward mechanism [48].

In our study, DIAB patients did not present with a faster functional decay until first visit as recorded by the ALSFRS‐R. Therefore, it does not seem that faster deterioration of the corticospinal pool was an explanation for the respiratory involvement found in ALS patients with DM. A possible local peripheral dysfunction would be more plausible. There are no relevant serial studies investigating phrenic nerve function in patients with DM, but a common subclinical lesion could be accepted as probable. Our results favor an association between phrenic nerve dysfunction and DM in ALS, with implications for respiratory outcome. Longitudinal studies should be performed, which would allow for a better comprehension of the impact of DM on neuronal and axonal function in patients with ALS, also addressing other muscles implicated in the respiratory mechanics, such as the accessory and paraspinal muscles. Additional studies are also needed to investigate the burden of cognitive dysfunction in ALS patients with DM. Unfortunately, we did not systematically evaluate cognitive changes in our population.

We conclude that respiratory function should be closely monitored in ALS patients with DM, as these patients present, even at first visit, significantly more severe respiratory involvement. Phrenic nerve studies should be considered. Early NIV is effective in compensating the respiratory dysfunction in these patients.

AUTHOR CONTRIBUTIONS

Susana Pinto: Methodology; validation; writing – original draft; writing – review and editing; investigation; formal analysis; data curation; software; conceptualization; resources. Marta Gromicho: Data curation; software; writing – review and editing. Michael Swash: Writing – review and editing; supervision.

FUNDING INFORMATION

Project Brainteaser–Bringing Artificial Intelligence Home for a Better Care of Amyotrophic Lateral Sclerosis and Multiple Sclerosis was funded by the European Union's Horizon 2020 research and innovation program under grant agreement GA101017598.

CONFLICT OF INTEREST STATEMENT

None of the authors has any conflict of interest to disclose. The authors alone are responsible for the content and writing of this article.

Pinto S, Oliveira Santos M, Gromicho M, Swash M, de Carvalho M. Impact of diabetes mellitus on the respiratory function of amyotrophic lateral sclerosis patients. Eur J Neurol. 2024;31:e16129. doi: 10.1111/ene.16129

DATA AVAILABILITY STATEMENT

The raw data are available from the first author upon reasonable request and upon discussion with the first and last authors.