Natural history of respiratory muscle strength in spinal muscular atrophy: a prospective national cohort study

Patients participated in a prospective clinical cohort study on SMA. We captured patient characteristics using standardized questionnaires and physical examinations, including motor function assessments and lung function tests, as described previously [8, 16, 17]. We used patient data obtained prior to participation in a clinical trial or treatment with SMN protein augmenting drugs (i.e., ‘treatment-naïve’). The local Medical Ethical Committee approved this study (09-307/NL29692.041.09) and informed consent was obtained from all participants and/or their parents in case of minors. The reporting of this study conforms to the STROBE statement [18]. Homozygous loss of SMN1 function and SMN2 copy number were determined using multiplex ligation-dependent probe amplification (MLPA; SALSA kit P021-B1-01, MRC-Holland) [16]. Using the SMA classification system, we distinguished different SMA types as described previously (Additional File 1) [2, 6, 17, 19].

We used respiratory muscle strength data of included patients collected during regular visits to the outpatient departments of pulmonology and the Center for Home Mechanical Ventilation at our hospital. Not all tests were performed at each visit. Data were not collected during hospital admissions, to prevent inclusion of measurements that are influenced by, for example, the presence of respiratory tract infections. For this study we used tests on expiratory strength [maximal expiratory pressure (PE_max), peak expiratory flow (PEF), peak cough flow (PCF)] and inspiratory strength [maximal inspiratory pressure (PI_max), sniff nasal inspiratory pressure (SNIP)]. All measurements were unassisted, i.e. not following manual compression or lung volume recruitment with frog-breathing, air stacking or mechanical insufflation–exsufflation.

PE_max and PI_max are non-invasive tests for the direct measurement of strength of expiratory and inspiratory muscles [15]. We measured PE_max and PI_max from total lung capacity and residual volume respectively, using the Geratherm Spirostik®. A nose-clip and flanged mouthpiece were used. Air leakage was prevented by a technician holding the lips. In some cases an oronasal mask was used. At least 5 repeated attempts were made. We recorded largest pressures and compared to the reference values provided by Wilson [20]. We calculated the PE_max/PI_max ratio to assess the relative impairment of expiratory versus inspiratory muscles [21, 22]. SNIP is nasal pressure measured during a maximal sniff. It is a simple test of inspiratory muscle strength. We measured SNIP in both nostrils using the Micro Medical MicroRPM®. Maximal nasal pressure during at least 5 sniffs, performed from Functional Residual Capacity, was compared to reference values [23, 24]. Finally we measured maximal flow during expiration and cough: PEF and PCF. We obtained PEF values from flow-volume curve data as the maximal expiratory flow achieved from forced expiration following maximal lung inspiration, using the Geratherm Spirostik® spirometer. We recorded the largest outcome from at least 3 qualitatively acceptable attempts. We reported absolute PEF values and standardized values [25, 26]. We measured PCF by maximal cough using both spirometry (Geratherm Spirostik®) and a peak flow meter (Assess®, PT-medical). We recorded the best outcome of 3 qualitatively acceptable attempts. Outcomes were reported as absolute values and compared to reference values [27, 28].

We measured all outcomes according to international guidelines [29], with patients in sitting position, without wearing corsets or braces. We used strong verbal encouragement and visual feedback to achieve maximal and reproducible test results. A resting period between tests prevented a significant influence of fatigability. All tests were performed by a small team of experienced professionals. We reported some outcomes as standardized values, i.e. as a percentage of the predicted value for age, height, weight, and sex. It is important to recognize that measuring height in SMA patients can be challenging. Tape-measured arm span was used preferably as a surrogate measure in patients unable to stand [8, 30].

We performed longitudinal analyses of PE_max, PI_max, PE_max/PI_max ratio, PEF and PCF. We used a cross-sectional analyses to assess the differences in SNIP outcomes between SMA types. A longitudinal analysis was hampered due to a too limited number of observations.

For the longitudinal analyses we used all available measurements and hypothesized progressive worsening of respiratory muscle strength over time, depending on SMA type. As it was unlikely that the longitudinal patterns were completely linear, we used non-linear analyses. We fitted smoothed B-spline models with 3 knots, in which polynomial continuous regression lines were computed in-between knots [31]. For PEF we additionally assessed the longitudinal pattern with a linear mixed-effects model (LMM). The model contained age, SMA type and an interaction term of these two predictors as fixed factors. Dependency in data due to repeated measures was accounted for by a random intercept per individual. A random slope for age was added to assess differences in rates of decline between patients (as a measure of disease heterogeneity or between-patient slope variability). We evaluated whether the rate of decline over age was significantly different between SMA types using a likelihood ratio test. For cross-sectional analysis of SNIP the first measurement of all patients was used. For the cross-sectional comparisons between SMA types we hypothesized that patients with milder SMA types would be less affected. As the assumptions of normality were met, a one-way ANOVA was used for comparisons between SMA types. A possible trend of increasing respiratory muscle strength with milder SMA types was assessed using the Jonckheere-Terpstra trend test.

We included 80 patients with genetically confirmed SMA types 1c–3b in this study. Ages at measurements ranged from 4.1 to 66.6 years. Baseline characteristics are shown in Table 1.

We analyzed 651 longitudinal measurements of PEF from 79 patients (Table 1). At baseline, PEF values differed significantly between SMA types (Fig. 1), i.e. 49%, 73%, 87% and 96% in SMA types 1c, 2a, 2b and 3a, respectively. The estimate for patients with SMA type 3b is unreliable, due to a limited number of observations (Table 1). PEF decline to values < 80% was observed in early childhood in SMA types 1c–2b, but not until adolescence or early adulthood in type 3a. In our linear analyses the average annual rates of decline did not differ significantly between SMA types (χ²(₄) = 6.2533, P = 0.181). PEF declined with 0.9%, 2.0%, 1.8%, 1.3% and 1.4% per year in SMA type 1c, 2a, 2b, 3a, and 3b respectively (model parameter estimates are shown in Additional File 2).

Non-linear analyses corroborate that PEF decline during early life is largely linear in most SMA types. In SMA type 2a this decline appears to be much faster during early childhood in comparison to children with type 2b. In adults with SMA types 2a, 2b and 3a we observed relative stabilization, although the data suggest that PEF decline can still occur during adulthood. Absolute values of PEF for the different SMA types are shown in Fig. 2.

We obtained 288 measurements from 61 patients. Longitudinal analyses are shown in Fig. 3, in which the important therapeutic thresholds of 270 L/min [indicating vulnerability to respiratory failure during otherwise trivial respiratory tract infections (RTIs)] and 160 L/min (indicating the boundary below which secretion clearance becomes ineffective) are marked [32].

PCF was lowest in SMA type 1c, with values < 160 L/min throughout life. After early childhood, patients with SMA type 2 reached values between 160 and 270 L/min, with clear differences between types 2a and 2b. Median PCF remained around 160 L/min in type 2a during adolescence and early adulthood, whereas in type 2b median PCF steadily increased until (early) adulthood. Patients with SMA type 3a had higher PCF values from earlier ages onwards in comparison to type 2b, but median values were still well below normal. The limited available data obtained from patients with type 3b indicate that even for these more mildly affected patients, PCF values may decrease in aging individuals.

We analyzed 586 measurements from 75 patients (Fig. 4), showing lower PE_max values from early childhood onwards in patients with SMA types 1c–3a compared to the reference population, where PE_max values are usually ≥ 80 cmH₂O during adulthood [20]. Patients with type 1c had severely lowered PE_max, without improvements with increasing age. PE_max in types 2a and 2b increased in adolescence to 40–50 cmH₂O. It is noteworthy that, despite limited data, all PE_max values from patients with SMA type 3b were < 80 cmH₂O and suggestive of a decline later in life.

We assessed PI_max longitudinally using 590 measurements from 76 patients (Fig. 5). Large intra- and inter-individual differences were present, in accordance with findings in the reference population [33]. Overall, PI_max was most affected in type 1c without improvements with increasing age. In patients with type 2a, PI_max increased to approximately 50–60 cmH₂O in adolescence. By contrast, patients with SMA type 2b reached PI_max values > 80 cmH₂O during adulthood. Patients with type 3a had a similar pattern, although in our cohort they did decline well below 80 cmH₂O from approximately 30 years onwards. The limited number of observations precludes definite conclusions for SMA type 3b.

We obtained 582 measurements from 75 patients. Figure 6 summarizes the longitudinal course, with a median ratio < 1 for all SMA types, except for a small number of older patients with SMA type 3a (but not type 3b).

We used the available SNIP data from 57 patients [median age: 12.9 years (IQR 9.9–29.0)]. SNIP was not statistically different between SMA types (F(4,52) = 2.219, P = 0.080). Median SNIP was 33, 44, 59, 58 and 55 cmH₂O in SMA types 1c, 2a, 2b, 3a, and 3b, respectively. We found a significant trend of increasing SNIP values with milder types (JT = 743, P = 0.0053). Importantly, virtually all SNIP outcomes were below 75 cmH₂O, which is considered the lower limit of normal (Fig. 7).