Acute Dyspnea After Use of Non-Invasive Ventilation in COPD Patients: The Deventilation Syndrome

Background: In daily routine, many COPD patients lament early onset augmented dyspnea following use of NIV (Deventilation Syndrome, DVS) as a negative side-effect. This is the rst study to this date to quantify and characterize DVS. This monocenter prospective observational study collected demographic, physiologic and symptomatic data from 67 in-patients with severe COPD Gold III-IV and chronic hypercapnic failure before, during and after use of an established NIV. During their inpatient follow-up, we examined patients during the rst hour after termination of nocturnal NIV. DVS was dened by the authors as an increase of ≥ 2 points on the Borg scale during the rst 30 minutes. We monitored cardiovascular and respiratory data and measured diaphragm excursion. Subjective dyspnea was documented by use of the Borg scale and questionnaires. In addition, respirator and demographic data were collected. an acute increase of dyspnea ≥ 2 points via Borg scale after termination of NIV. A main key cause of DVS may be breath rate acceleration and dynamic hyperination after the use of NIV.


Introduction
Non-invasive positive pressure ventilation (NPPV, NIV) is a widespread therapeutic option for patients with chronic hypercapnic failure due to COPD [1,2] Effectively reducing hypercapnia by high-intensity ventilation can not only lead to improvement of health-related quality of life and lung function status, but has shown to reduce hospital re-admission and prolong long-term survival [1]. Persistent improvements of dyspnea in the daytime have also been documented [3,4].
Aside many positive aspects, there are also negative side effects of NIV. Most commonly described are dryness of the upper airways, abdominal bloating or sleep fragmentation [5,6], but also multifacetted fear of therapy [7].
This study now describes a further, highly burdening side effect: early onset acute dyspnea following the use of NIV -augmented in the morning, but also after use during the day.
During ongoing NIV, patients experience relief of dyspnea and can breathe without effort. In some patients, termination of NIV within minutes leads to increasing dyspnea and discomfort. This is accompanied by tachypnea with rapid, shallow breathing, increased heart rate, the need to sit upright (coachman's seat) and the feeling of restlessness, sometimes even leading to panic attacks and fear of use of NIV. This state is hereby de ned as the Deventilation Syndrome (DVS).
The nomenclature "Deventilation Syndrome" is rst found in a publication by Adler et al. [8], describing morning dyspnea mainly due to ventilator desynchronization in sleep medicine.
The novel use of this term has been adapted to describe augmented post-NIV dyspnea in an adequate and su cient ventilatory situation.
COPD patients with long-term home ventilation (LT-NIV) were surveyed during a routine in-patient exam and additionally monitored in the rst hour of post nocturnal use of NIV -thus establishing not only occurrence of DVS, but also possible physiological mechanisms. This is the rst study to this date to de ne and characterize DVS in patients with COPD and LT-NIV.
Despite high prevalence and symptom burden, DVS so far remains widely unaddressed.

Design
In a monocenter prospective observational study demographic, physiological and subjective data was collected from in-patients with severe COPD III-IV and chronic hypercapnic failure before, during and after use of an already established NIV.
Extended routine examinations included cardio-respiratory monitoring during and one hour after the nocturnal use of NIV, including ultrasonic diaphragm measurements. Subjective dyspnea was documented before and after use of NIV by use of the Borg scale and questionnaires were supplemented.
Being an observational study to document the Deventilation Syndrome, no study endpoints were de ned.
The authors refrained from power calculation due to the observational character of the study.

Patients
Between November 2016 and January 2019, 71 patients with severe COPD (Gold III°-IV°) and established long-term home NIV were included in the study. Su cient data was collected from 67 patients. Inclusion criteria were: men and women with stable COPD disease, GOLD stage III-IV, regular use (≥ 4h/d) of NIV (prescription ≥ 1 year) and age > 18 years. Study participation required written consent and patients had to be able to follow the requirements of the project. The study protocol was granted permission by the ethics review board of the University of Heidelberg, Germany and all patients gave their written consent.
During the study any medically necessary concomitant therapy was allowed. Intercurrent diseases were treated according to the clinic standard.
Exclusion criteria were: Acute pulmonary impairment (e.g. pneumonia), other acute diseases such as acute pulmonary artery embolism, haemoptysis, internal bleeding, pneumothorax. In addition, serious neurological disorders (e.g. apoplex) or cardiovascular diseases with hemodynamic instability (e.g. heart failure NYHA≥ III, myocardial infarction < 1 month ago) were excluded, as also newly occurring hypoxemia or worsening of hypercapnia ≥ 20% compared to previous values. In addition, alcohol/drug/drug abuse made participation in the study impossible.

Deventilation Syndrome (DVS)
DVS is determined by dyspnea -a highly subjective symptom. Objecti cation is best done by use of the modi ed Borg scale [9] The minimal clinical importance difference for Borg is 1 point [10]. Being a novel syndrome, there is no de nition of DVS to this date. For this study and further use of this terminology, DVS was de ned by an increase of subjective dyspnea by ≥ 2 points on the Borg scale during the rst 30 minutes after NIV termination -thus stressing the symptomatic and chronologic context after use of NIV.

Survey examination timeline
Baseline (T0, day of admittance): anamnesis, PE, questionnaires, BGA w/wo NIV, pulmonary function tests, spO 2 , breath and heart rate, blood pressure, ultrasound measurement of diaphragm excursion Tn (following morning, with NIV): dyspnea by Borg scale, spO 2 , tCO 2 , breath and heart rate, blood pressure T1 (10 min post NIV): dyspnea by Borg scale (and every 10 min for the following hour), BGA, spO 2 , tCO 2 , breath and heart rate, blood pressure, ultrasound measurement of diaphragm excursion T2 (60 min post NIV): dyspnea by Borg scale, BGA, spO 2 , tCO 2 , breath and heart rate, blood pressure Pulmonary Function Diagnostics These were carried out according to the internal standards of the lung function department of the University Thoraxklinik Heidelberg, using MasterScreenBody and PFT Body + Diff. by Care Fusion (Jaeger).
Reference values were calculated according to GLI for spirometric parameters and TLCO [11,12], while ECSC was referenced for body plethysmography [13].

Respiratory muscle strength:
Exhaustion of respiratory muscle function was measured non-invasively by maximal inspiratory mouth pressure (Pimax), re ecting respiratory capacity. The respiratory load was documented by measurement of occlusion pressures (P0.1) [14].

Ultrasonic diaphragm measurement
Measurements of diaphragm excursion were performed using the Philips Ultrasound Sparq with an abdominal transducer C 6-2. Patients were examined during spontaneous breathing on admission day (T0) and the following morning after termination of NIV (T1). This was done in a dorsal lying position, the torso elevated at approx. 45°.The right diaphragm was located in B-Mode and then excursion measured in M-Mode subcostal, mid clavicular line [15,16].

Measurement of carbon dioxide values and blood gas analysis
Carbon dioxide values were collected by two methods: continuous transcutaneous measurement and capillary blood draw (blood gas analysis, BGA), both examined on the earlobe. BGA was assessed by arterialized capillary ear-lobe puncture earlobe skin [17]. The ABL 800 ex from Radiometer GmbH was used as blood gas analyzer. Transcutaneous CO 2 measurement was performed using TOSCA transcutaneous monitor TCM 4, Radiometer.

Monitoring
Nightly monitoring of patients undergoing long term home NIV is an essential part of NIV surveillance, re ecting the respiratory situation and e ciency of ventilation during sleep. This standardized procedure was performed during nocturnal use and extended one hour after termination of NIV on our ward using Philips IntelliVue MX550 monitors.

Questionnaires
Borg perception of exertion CR10 Scale Acquired by Modi ed Borg Scale (MBS) with 10-point dyspnea assessment [9]. An increase of ≥ 2 points after NIV termination was de ned by the authors as relevant for DVS.

mMRC: Modi ed Medical Research Council
Score from 0 ("never breathless, except during intense exertion") to 4 ("too short of breath to leave the house or to dress and undress alone") [19].

Statistical analysis
Baseline characteristics are presented for the whole sample and strati ed by group (DVS vs nDVS) by mean and standard deviation (SD) or median and interquartile range (IQR) or frequencies (n, %) as appropriate. Differences between groups were investigated by means of independent t-test, Mann-Whitney U test or chi-square test retrospectively. Continuous variables were explored graphically to justify parametric analysis.
Longitudinal models were applied to test for time x group effects for outcomes measured over time. Except for BORG (ordinal scale), a linear regression model with repeated measures statement was applied for each outcome with the explaining variables time, group, and interaction term. Group and time effects with con dence intervals and p-value were presented based on the contrast between groups at each visit and contrasts between the visits for each group.
For Borg, the total score was analyzed descriptively by median (IQR) at the four measurement points in time for each group (DVS and nDVS).
To analyze the relationship of Borg and FEV1, Spearman correlation coe cients were calculated. This is an exploratory analysis so that p-values are interpreted descriptively. A p-value < 0.05 was considered statistically signi cant. All analyses were performed in SAS version 9.4.

Baseline Characteristics
Between November 2016 and January 2019, 71 patients were included in the study. One patient was excluded due to double enrollment, three patients did not complete the Borg scale, therefore 67 patients were further evaluated. 39 patients were retrospectively diagnosed with DVS, 28 did not present DVS symptoms.
Enrollment included overall 28 males and 39 females (corresponding to 42% and 58% respectively). The mean (SD) age of all patients was 66.1 ± 6.7 years. Comorbidities were widespread, but severe and/or acute diseases were excluded according to the exclusion criteria of the study protocol.
No serious complications occurred during the study, e.g. acute exacerbation of the disease. All patients were compliant with both participation in the study and with their NIV therapy. However, 5 patients did not wish to perform lung function tests. In all, 13 patients were unable to undergo or complete body plethysmography due to their physical constitution and extremely high lung in ation, one further patient had non legible RV rates. Interestingly, 11 of these 13 patients were retrospectively diagnosed with DVS.

Ventilatory setting
Ventilatory settings (VS) were documented ( Table 1). The majority of patients was ventilated in ST mode (spontaneous/timed mode), 2 patients of the cohort used an aPCV mode (assisted pressure-controlled ventilation).

Lung function analysis
Subgroup analysis of lung function measurements showed higher pulmonary impairment in patients with DVS and a clear trend to higher lung in ation. FEV1 was lower than in nDVS (0.6 vs 0.8l, p 0.016), corresponding to a mean FEV1 of 25% (GOLD IV) vs nDVS FEV1 mean 33% (GOLD III). Lung in ation was elevated in DVS, re ected by RV (6.1 vs 5.4l, p 0.10) and respiratory muscle strength via Pimax(RV) was reduced in the DVS group compared to nDVS (4.0 vs 5.0 kPa, p 0.06, though both were not statistically signi cant due to a high drop-out rate in body plethysmography (insu cient maneuver).

Carbon dioxide levels
When analyzing subgroup data, focus was placed on transcutaneous CO 2 measurements. As there was no transcutaneous T0 data, capillary CO 2 measurements were alternatively respected.
Patients with DVS had higher CO 2 levels than nDVS patients at all times of measurement (Table 3).
Hypercapnia was decreased during NIV in both subgroups, but remained higher in DVS (49 vs 45mmHg, p 0.05). This effect was even clearer after NIV termination at T1 (54 vs 48mmHg p < 0.01) and T2 (53 vs 47mmHg, p < 0.01).  Oxygen There were no signi cant differences in oxygen saturation, both as independent variables at the different points in time and in a group as a dependent variable.

Ultrasonic measurement of diaphragm excursion
The ultrasonic examination of the diaphragm was performed in all patients on day of admittance (T0) and the next morning after NIV termination (T1). In the overall group, T0 mean diaphragm excursion (± SD) was 1269 ± 559 mm, at T1 1069 ± 456 mm. Excursion difference between T0 and T1 (mean − 200mm) was statistically signi cant (p 0.03).

Borg scale CR10
67 patients correctly lled out the Borg scale.
For this trial, an increase of ≥ 2 points on the Borg scale within 30 minutes after NIV termination was de ned as Deventilation Syndrome, based on the patient's subjective dyspnea.
This criterion was met by 39 out of 67 patients (Table 5), corresponding to a share of 58.2%. The majority of patients (36/39) experienced dyspnea immediately after NIV termination (≤ 10minutes), only 3 patients showed symptoms between 10-30 minutes.

Table 5 Distribution of DVS and nDVS
In the DVS group, baseline (T0) median (IQR) Borg score was 3.0 (3.0) points and dropped to 1.0 (3.0) points undergoing NIV (Tn). After NIV termination, Borg value increased to median of 5.0 (4.0) points during the rst 10 minutes, over 90% of the DVS patients experienced greater respiratory distress in this rst time segment. Within the next hour, median Borg values dropped continuously and settled after 60 min at 3.0 (3.0) points -thus meeting baseline levels (Fig. 2).
In the nDVS group, baseline (T0) median (IQR) Borg score was 1.0 (3.5) points. During NIV (Tn) there was a decrease to 1.0 (2.0) points. Stopping the use of NIV led to a very slight increase to a median of 1.0 (3.0) points 10 minutes after, then rapidly sinking back to 1.0 (2.5) points. This value remained constant for the rest of the monitoring period. It is noteworthy that this decrease was much faster than in the DVS group and Borg even remained lower than spontaneous breathing on the day of admission (Fig. 2).

Borg Scale and FEV1
As to be expected, there was a high correlation of Borg scale rankings and the severity of pulmonary impairment, measured by FEV1. Lower FEV1 levels were associated with more dyspnea (p < 0.01).

CAT
The CAT questionnaire documented a higher mean score in DVS patients than nDVS (25.5 ± 6.8 points vs 19.8 ± 7.5 points, maximum score 40 points), re ecting lower FEV1 levels and higher disease burden. This effect was signi cant p < 0.01).

Discussion
Acute dyspnea after the use of non-invasive ventilation (NIV) in a long-term setting is a common and stressful symptom in COPD patients. This is the rst trial documenting prevalence and physiology of this symptom, hereby de ned as Deventilation Syndrome (DVS). Despite an abundance of publications on NIV in COPD, many also addressing negative side-effects, DVS has remained widely unaddressed.
DVS was identi ed in 58% of COPD patients with chronic hypercapnic failure in a routine in-patient setting, thus emphasizing the relevance of the subject. Ventilator settings were effective in both groups (DVS and nDVS), leading to signi cant decrease of CO 2 levels and documented reduction of dyspnea per Borg scale undergoing NIV. Both groups were fully adapted to the regular use of NIV and showed no signs of discomfort or asynchronicity undergoing therapy. Nightly cardiovascular and pulmonary monitoring showed no signs of relevant sleep disordered breathing.
Dyspnea is a complex symptom and still not fully understood, though its role as a predictive marker in chronic lung diseases has been acknowledged [20][21][22]. It is highly subjective in perception and does not necessarily correlate with speci c tests (e.g. spirometry, blood gas analysis), but dyspnea can -and should -be measured. Underlying mechanisms and qualities of dyspnea can be categorized in: elevated work or effort in breathing (impaired respiratory muscles), tightness of the chest (bronchoconstriction) and intensity of air hunger (increased inspiratory drive, desensitized pulmonary mechanoreceptors) [22]. Following these main mechanisms can help understanding causes and thus provide therapeutic concepts in DVS.
Ultrasonic measurement of the diaphragm showed signi cant (pre NIV 1269mm, post-NIV 1069mm, p 0.03) movement impairment after the use of NIV in the overall group, possibly expressing transient reduction of diaphragm muscle strength. There was no signi cance between each subgroup, though the differences of diaphragm movement pre and post NIV do strike as noteworthy (nDVS 260mm vs DVS 157mm). Ventilator induced diaphragm dysfunction (VIDD) is a well-known entity that has not only been addressed in invasive-ventilator ICU settings [23][24][25], but also has been seen in NIV in COPD patients [26], though its role here is yet to be established. DVS shows an impact after use of NIV: in DVS, RR was higher than nDVS patients (T 10min : 20.1 vs 18.1 bpm, p 0.04) and remained elevated during the hour of monitoring (T 60min : 20.4 vs 18.9 bpm, p 0.1, though most DVS patients were subjectively recovered by Borg scale 60 minutes after NIV termination in this study. RR after use of NIV in DVS patients was higher than at T0 (baseline) although not statistically signi cant, so time adjusted RR change should be acknowledged. During the short period of 10 minutes post NIV, DVS patients showed a mean RR increase of 4.3 bpm, in nDVS 2.8 bpm.
Potential reasons for acute RR increase could be reduced diaphragm strength or elevated inspiratory drive due to higher CO 2 levels in the DVS group (tCO 2 T 10min postNIV 54 vs 48 mmHg, p < 0.01).
Also, higher overall bronchial obstruction in DVS (FEV1 25% vs 32.5% nDVS, p 0.01), furthermore augmented obstruction in the early morning hours (circadian, vagal stimulation, decreased medication levels), increases breathing muscle activity and effort and can trigger dynamic hyperin ation in already severely in ated lungs (RV DVS 6.1 vs nDVS 5.4 liters, p 0.1), thus circling into rapid shallow breathing (RSB). RSB and impaired diaphragm function have been identi ed as predictors for weaning failure [27], re ecting relevant breathing muscle exhaustion. A possible negative in uence of DVS on long-term prognosis of COPD patients with NIV is yet to be surveyed.
It is also noteworthy that many DVS patients expressed fear of NIV termination in the morning due to experienced dyspnea in the past, posing a further possible cause for RR acceleration Oxygen levels at any time also did not differ in both groups and thereby remain disregarded.
There are limitations in this trial to be discussed. First, this was a monocenter study of 67 patients, further measurements need to be done in a multicenter setting. Second, assessment of diaphragm function was performed only by ultrasonic diaphragm excursion (DE) and only in two examinations, showing high standard deviation. Initially, DE was only thought an addendum to other focus topics -it now evolved to possibly one of the key features in DVS. In the future, more detailed data (e.g. diaphragm thickening index, repeated measurements) will be needed. Third, there was no guideline in the protocol concerning medication intake. This was intentionally put in the hands of each patient to decide by need, hence heterogenous treatment patterns. Subsequent trials should standardize this procedure. And fourth, monitoring was only continued for one hour during which some patients did not return to baseline level in terms of dyspnea, respiratory rate and CO 2 levels. So, full duration of DVS has not yet been captured.

Conclusion
In conclusion, DVS is a frequent, but to this day insu ciently regarded side effect of LTH-NIV. The recognition of DVS is of high clinical relevance for health care givers and patients. This is the rst trial documenting not only prevalence, but also relevant patient data leading to concepts of factors that presuppose DVS, by this stepping toward prevention and therapy of this often tormenting symptom. Further research is required to substantiate DVS and establish alleviating therapies. First authorship is declared by the submitting author MS. MS concepted study design, acquired participants, collected and interpreted data. SI and GI collected and analyzed substantial data in COPD patients including questionnaires and diaphragm ultrasound. MK analyzed and statistically approved all collected data. FH and FT analyzed and interpreted lung function and demographic data. All authors read and approved the nal manuscript.