Rationale and Design of a Mechanistic Clinical Trial of JAK inhibition to Prevent Ventilator-Induced Diaphragm Dysfunction

Abstract

Introduction:

Ventilator-induced diaphragm dysfunction (VIDD) is an important phenomenon that has been repeatedly demonstrated in experimental and clinical models of mechanical ventilation. Even a few hours of MV initiates signaling cascades that result in, first, reduced specific force, and later, atrophy of diaphragm muscle fibers. This severe, progressive weakness of the critical ventilatory muscle results in increased duration of MV and thus increased MV-associated complications/deaths. A drug that could prevent VIDD would likely have a major positive impact on intensive care unit outcomes. We identified the JAK/STAT pathway as important in VIDD and then demonstrated that JAK inhibition prevents VIDD in rats. We subsequently developed a clinical model of VIDD demonstrating reduced contractile force of isolated diaphragm fibers harvested after ~7 vs ~1 hour of MV during a thoracic surgical procedure.

Materials and Methods:

The NIH-funded clinical trial that has been initiated is a prospective, placebo controlled trial: subjects undergoing esophagectomy are randomized to receive 6 preoperative doses of the FDA-approved JAK inhibitor Tofacitinib (commonly used for rheumatoid arthritis) vs. placebo. The primary outcome variable will be the difference in the reduction that occurs in force generation of diaphragm single muscle fibers (normalized to their cross-sectional area), in the Tofacitinib vs. placebo subjects, over 6 hours of MV.

Discussion:

This trial represents a first-in-human, mechanistic clinical trial of a drug to prevent VIDD. It will provide proof-of-concept in human subjects whether JAK inhibition prevents clinical VIDD, and if successful, will support an ICU-based clinical trial that would determine whether JAK inhibition impacts clinical outcome variables such as duration of MV and mortality.

Introduction

Mechanical ventilation is a life-saving measure for patients who are incapable of independently ventilating and/or oxygenating. An estimated one million people per year receive ICU-based MV in the United States alone (1). Unfortunately, even brief periods of MV (>12h) result in clinical diaphragm weakness (“ventilator-induced diaphragm dysfunction” – VIDD) that renders it difficult to wean patients from the ventilator. VIDD is characterized by an early loss of contractile force of diaphragm muscle fibers, and later frank atrophy of fibers, which combine to dramatically reduce the muscle’s ability to generate inspiratory force (2–5).

Prolonged ventilator dependence, which often results at least in part from VIDD, occurs in up to 34% of those ventilated (6–11), and it ranks third in total inpatient charges among Medicare diagnostic groups (12). As just one example of the many complications associated with prolonged MV, ventilator-associated pneumonia (13) occurs in up to 27% of ventilated patients (13, 14), its incidence increases with time on the ventilator (15), and its mortality has been estimated to be between 33 and 50% (16).

Evidence that VIDD is indeed a major contributor to failure to wean from MV is abundant. First, several clinical trials of patients undergoing attempted weaning show that low diaphragm strength is associated with poor weaning outcomes (17, 18). Secondly, patients who had failed to wean eventually did wean successfully after diaphragm force-generating capacity (Pdimax) improved (19, 20). Thirdly, clinical trials investigating the effect of inspiratory muscle strength training (IMST) in patients who had failed to wean show that the resulting improvement in maximal inspiratory pressure reduces the average weaning period (in one study) or increases the rate of successful weaning (21, 22).

In an effort to identify the central pathogenic events underlying VIDD, we and others reported that mitochondrial oxidative stress (23) is induced in MV diaphragm (23, 24). In the diaphragm of human subjects, MOS appears to be a key upstream inducer of proteolysis, which underlies diaphragm myofiber atrophy (23). MOS can also impact the earlier-occurring, specific force component of VIDD in at least two ways. First, MOS generates free radicals that oxidize muscle proteins, altering their structure and function including actin-myosin cross-bridge kinetics and/or the calcium sensitivity of myofilaments (25). Second, MOS induces a metabolic switch – reducing mitochondrial oxidative phosphorylation and increasing glycolysis (26) – which may lead to reduced energy supply to the muscle and thus muscle weakness on that basis.

Critical as the theoretical basis of this clinical trial, we and others have demonstrated that the JAK/STAT pathway plays a crucial role in the development of VIDD as a result of its influence on mitochondrial function and the induction of MOS (23, 27). Both JAK and STAT are significantly phosphorylated/activated in human diaphragm muscle after MV (23, 27). Most importantly, JAK inhibition prevents VIDD in mechanically ventilated rats (27, 28) (Figure 1). While evidence suggests that the relationship between JAK/STAT activation and mitochondrial ROS production is likely bidirectional (i.e., mutual induction of one another), its role in this scenario appears to be at least substantially in the direction of JAK/STAT inducing ROS.

Figure 1.

JAK inhibition prevents the reduction of muscle-specific diaphragm force that occurs in mechanically ventilated rats. A) MV induces STAT3 phosphorylation and JAK inhibitor R545 blocks this activation. Control and MV (18 h) rat diaphragm muscles were subjected to Western blot analysis (n=4 rats per group). B) Treatment with JAK inhibitor prevents MV-induced contractile dysfunction. Force frequency relationship from stimulated, ex-vivo diaphragm strips from controls (n=10) and MV rats treated with JAK inhibitor (n=8) or vehicle (n=9) (adapted from ref 27).

We designed a human model of VIDD to allow us to test whether JAK inhibition prevents human VIDD. Our model exploits the surgical procedure esophagectomy – the longest commonly performed operation that provides access to biopsy the diaphragm (and a control muscle) at the start (~1.0 h of MV) and the end (~7.0 h of MV) of the operation. We hypothesized that 6 h was a reasonable time over which to expect the diaphragm weakness that characterizes VIDD (though not the atrophy) to develop. We have indeed demonstrated using this clinical model that STAT3 (in the JAK/STAT pathway) is significantly phosphorylated (activated), and that the maximal force generated by single, isolated diaphragm muscle fibers significantly falls, between 1 h and 7 h of MV in human diaphragm (Figures 2,3).

Figure 2.

Human diaphragm muscle fibers, similar to rat, demonstrate activation/phosphorylation of STAT3 after 7 hours of mechanical ventilation. A) Western blots from 5 representative subjects demonstrate similar STAT3 levels at 1 and 7 h, but increased pSTAT3 at 7 h. B) The pSTAT3 to STAT3 ratio is significantly elevated at 7 h vs. 1 h (n=14, *p=0.032)

Figure 3.

Maximal force generation of isolated, skinned single human diaphragm fibers is reduced at 7 h MV vs. 1 h MV (p=0.024). Each point represents a single fiber, with bars showing the range of forces measured. n=39 fibers from 6 subjects. See text for detailed methods.

This NIH-funded clinical trial (CT.gov #NCT03681275) is the logical extension of this preliminary work and will test the hypothesis that JAK inhibition in human subjects will prevent both the molecular/biochemical changes within the diaphragm and the diaphragm muscle weakness characteristic of VIDD. The trial employs our novel clinical model of MV during esophagectomy to determine if treatment with the FDA-approved JAK inhibitor Tofacitinib will prevent the development of VIDD over 6 hours of MV. The study is a prospective, randomized, placebo-controlled clinical trial – with physiological and mechanistic outcome variables. Its success would provide the basis for proceeding with a subsequent clinical trial of JAK inhibition in ventilated ICU patients, with clinical outcome variables such as duration of MV and mortality rate.

Materials and Methods

Objectives

The primary hypothesis tested is that JAK inhibition with Tofacitinib will prevent VIDD (defined as reduction maximal isometric force of isolated diaphragm myofibers) and will prevent the molecular changes known to underlie VIDD, over 6 hours of MV in human subjects.

Primary Outcome

Reduction by the study drug in the MV-induced fall in the maximal isometric force of isolated, single diaphragm muscle fibers between diaphragm biopsies taken from human subjects after 1 h and 7 h of MV (Figure 3). A mean 30% reduction in the fall in contractile force due to MV would be considered a clinically sufficient reduction and encourage us to subsequently proceed toward an ICU-based clinical trial of JAK inhibition to prevent VIDD.

Secondary Outcomes

1. Reduction by Tofacitinib in the MV-induced elevation of the phosphorylation (activation) level of STAT3 in diaphragm muscle (between 1 and 7 h of MV; the pSTAT3:STAT3 ratio).

2. Reduction by Tofacitinib in the MV-induced elevation in several measure of MOS in diaphragm muscle (between 1 and 7 h of MV).

Additional Outcomes

Reduction by Tofacitinib in the MV-induced alteration of several other molecular and biochemical measures in diaphragm muscle (between 0.5 h and 6 h of MV) reflecting pathogenetic events known to underlie VIDD.

Biochemical/Molecular markers that will be measured: markers of muscle protein degradation (Atrogin, MuRF1, ubiquitination) and biochemical assays of proteosomal (muscle degradatory) activity; assays of mitochondrial function (ATP levels, citrate synthase activity, succinate dehydrogenase activity, mitochondrial respiration rate); mitochondrial oxidative stress measured by direct immunostaining for free radicals/superoxide (dihydroethidium and mitoSox staining), biochemical assays of H2O2 levels, and detection of the oxidized proteins by western blot analysis.

Study Design

The study is a parallel group design, randomized, placebo-controlled, double-blind clinical trial. 56 subjects, age 15 and older, preparing to undergo esophagectomy for esophageal cancer or benign disease will be recruited in the clinics of the thoracic surgeons at Stanford Medical Center. Participants are being approached for possible enrollment based upon the following inclusion/exclusion criteria. Inclusion Criteria

Age 15 or older

Undergoing esophagectomy by a surgical approach that will provide access to the planned muscle biopsies

Exclusion Criteria

Based upon potential Tofacitinib toxicities/drug interactions:

• ALT, AST, total bilirubin, or alkaline phosphatase level above 150% of normal (reducing Tofacitinib excretion)

• Creatinine level above 2.0 (reducing Tofacitinib excretion)

• Currently taking systemic antifungal medication (Tofacitinib levels are increased by some antifungals)

• Currently taking systemic immunosuppressive medications (to avoid marked immunosuppression; including corticosteroids, which may impact muscle structure/function)

• History of untreated TB, or a positive PPD or positive quantiferon test which has not been treated (since Tofacitinib can re-activate TB))

Based upon potential but unlikely impact of small diaphragm biopsies on respiratory muscle function:

• FEV1 or FVC under 70% predicted, or DLCO under 50% predicted (if measured)

• Neuromuscular disease that might compromise diaphragm function

Based upon general considerations:

• Pregnancy

• >5% weight loss over the last 6 months (which may indicate cachexia, which can be associated with JAK/STAT pathway activation and impact muscle structure/function)

Choice of Experimental Drug

As noted above, both our prior animal work and preliminary data in humans showed that STAT3 (in the JAK/STAT network) is markedly activated by MV. The JAK inhibitor, Tofacitinib, is an oral agent in common clinical use for patients with rheumatoid arthritis (RA), based upon data from 40 phase I-III clinical trials. It is a synthetic molecule that selectively inhibits JAK family members, with rank order potency JAK1>JAK3>JAK2>Tyk2. Numerous cytokines signal via JAK pathways, and inhibition of JAK activation attenuates these pro-inflammatory signals, in part through prevention of the activation of the STAT family of transcription factors. Only JAK 2, which is less inhibited by Tofacitinib than JAKs 1 and 3, subserves substantial effects outside of inflammatory cascades (e.g., JAK2 controls signaling for the hematologic growth factors erythropoietin and GMCSF).

While chronic use of Tofacitinib is rarely associated with adverse events related to reduced cell-mediated immunity, one would not expect such side-effects with the short term (~52 h) duration of drug administration proposed in our study. Hematologic adverse events have not occurred in significant number even with chronic use. The rate of adverse events in randomized clinical trials of the drug in RA was only 199 events per 100 patient-years of therapy. Even infectious events occurred in only 0.8% of subjects treated in long-term trials (29). In a randomized trial of Tofacitinib begun immediately after renal transplantation at triple the dose proposed in our study, there was no increase in surgical complications or survival in the Tofacitinib group (30). A recent trial of Tofacitinib in hospitalized patients with COVID-19 pneumonia, 24% of whom went on to non-invasive or invasive MV, demonstrated a reduction in respiratory failure and death in the Tofacitinib arm and no important difference in the incidence of adverse events (31). All of this information bodes well for the drug’s safety in our perioperative experimental model, as well as for potential use of the drug for a relatively circumscribed time-course of days, early after institution of MV, to try to prevent weaning failure and prolonged MV in ICU patients.

Oral Tofacitinib is rapidly absorbed (Tmax 0.5–1h) and eliminated (T1/2 of 3h). Steady state levels are reached after 24–48 h of the 5 mg bid dose used in rheumatoid arthritis and in our study. To be certain to reach steady state levels in our experimental subjects, we will administer five, 5 mg doses bid preoperatively over 48 h, and one additional dose intraoperatively via the nasogastric tube (at the ~52nd h), to be certain that therapeutic levels persist throughout the approximately 6 hour duration of MV that subjects will undergo.

Anesthetic

The anesthetic technique for esophagectomy will only allow utilization of neuromuscular blockers for endotracheal intubation at the start of the procedure, and not later in the procedure. Further, only cisatracurium will be allowed, which has been shown to have minimal detrimental effect on diaphragm function, in contrast to rocuronium (32, 33). The remainder of the anesthetic will be via narcotics alone. The prophylactic antibiotic used will be cefoxitin, which provides appropriate coverage for esophagectomy, given the known effect of several other antibiotics on neuromuscular function.

Experimental Drug / Placebo Administration

Consenting subjects will be randomized 1:1 to receive either Tofacitinib or placebo prior to and at the initiation of MV for their esophagectomy. Subjects and investigators will be blinded to the assignment (double-blind). The subjects will receive Tofacitinib or placebo according to the following schedule:

• 5 mg po q 12 h the 2^nd day prior to surgery

• 5 mg po q 12 h the 1^st day prior to surgery

• 5 mg po at home on the day of surgery, approximately 2.5 h prior to surgery

• 5 mg via nasogastric tube immediately after the start of MV

STAT3 phosphorylation levels will preliminarily be examined in diaphragm biopsies from 7 h vs 1 h of MV after the initial 10 subjects have been enrolled. If there is not significant reduction in STAT3 phosporylation levels seen in the experimental drug group vs the placebo group, then we will request an amendment from the Stanford IRB to increase each of our Tofacitinib doses to 10 mg, the higher of the 2 doses commonly employed in RA.

The entire study timeline, including experimental drug/placebo administration is shown in Figure 4

Figure 4.

Study timeline: includes drug administration schedule, surgery and diaphragm/serratus biopsies, and follow-up.

Sample Acquisition

In consenting subjects, blood will be drawn preoperatively and at the time of the second diaphragm biopsy, in order to measure systemic JAK/STAT3 activation/phosphorylation in mononuclear cells lysates at baseline and after MV.

Biopsies of diaphragm and serratus anterior (a non-respiratory muscle control that is available through the incisions made for esophagectomy) will each be taken at 2 time points during the esophagectomy procedure: 1) shortly after opening the abdomen ~1h after institution of MV; 2) immediately prior to closing the chest after ~7h of MV (Figure 5)

Figure 5.

Schematic of the clinical model used for the trial. Diaphragm and Serratus Anterior biopsies are harvested after 1 hour and 7 hours of mechanical ventilation while the patient/subject undergoes an esophagectomy. The diaphragm fibers (and control serratus muscle fibers) are studied for changes in STAT3 activation and for the development of diaphragm weakness in ex vivo study of single, isolated, skinned muscle fibers.

The diaphragm biopsies will be from the right, anterolateral, costal portion of the diaphragm in a region free of major phrenic nerve branches. They will each be 7 × 5 mm biopsies, and the 7 h biopsy will be taken from a different bundle of the radially-oriented diaphragm fibers than the 1 h biopsy, to be certain not to be harvesting from a bundle that might have been injured by the first biopsy. The serratus anterior biopsies will be taken from opposite sides of a 2 cm chest skin incision that runs parallel to the direction of the fibers, again in order to assure that the 2 biopsies are from different fiber bundles.

Sample Processing

The central approximately 50 percent of each biopsy sample is separated by sharp dissection, in the direction of the muscle fibers, to be prepared for single fiber force studies and mitochondrial isolation. The remaining 50 % of each biopsy (from the periphery of the samples) is snap-frozen and used for the biochemical/molecular studies.

Safety Considerations

Careful monitoring is being undertaken for adverse events that may be due to the experimental drug or to the biopsy procedures. Specific risks that may potentially be associated with the drug include an increase in infectious or wound/anastomotic healing complications following the operation. Specific risks that may be potentially related to diaphragm biopsies include increased respiratory complications such as pneumonia, atelectasis, re-intubation, longer duration of mechanical ventilation, pleural effusion, and the occurrence of diaphragmatic hernia.

The surgeons and study coordinator are monitoring for adverse events daily postoperatively, including daily chest x-rays, until discharge from the hospital on approximately postoperative day 8, and again at the time of the first postoperative visit one month postoperatively and at the standard next follow-up visit six months postoperatively. Adverse events are also being monitored by an Independent Medical Monitor (IMM) and an independent Data Safety Monitor Board (DSMB). The DSMB meets at least every 6 months and reviews adverse events and unmasked data. It will also be convened urgently within 1 week if 4 serious adverse events (SAE) felt to be either related or likely related to the study drug occur or if any member requests a meeting after reviewing any of the SAEs electronically. The study will be halted if the DSMB determines that there is a statistically significant, and greater than 20%, increase in any SAE or the total sum of AEs between the experimental and control groups, or between all of the study patients and historical esophagectomy controls. The study will also be halted pending further evaluation if there is a single esophageal hernia felt related or likely related to the diaphragm biopsies. There are no predetermined early stopping points outside of safety issues.

Single Fiber Muscle Force Measurement

The technique of single muscle fiber force measurement will not be described in detail here but is being carried out largely according to the protocol of Cooke et al (34). In brief, the fibers are chemically skinned by placement for 24 h at 4 °C in “Rigor buffer” containing glycerol. Single fibers are then dissected manually under magnification and mounted on a single fiber force transduction apparatus (Aurora Scientific). The sarcomere length of the fibers is set at 2.6 um, which is commonly accepted as the optimal length for contractility of human muscle fibers. Fibers are activated, and isometric maximal force measurements obtained, using a “calcium-jump” protocol, at 15 °C. Maximal active isometric force (Po) is calculated as the difference between the total force when activated and the resting tension. Measurement of fiber length and diameter allows calculation of cross-sectional area (CSA), which is required to ultimately calculate specific force (force/CSA). Fibers with visible tears and those in which a consistent resting sarcomere length of 2.6 cannot be established are excluded from the analysis. Twenty fibers per muscle sample are evaluated. After force measurements, each fiber is placed in SDS sample buffer to determine the myosin heavy chain (MyHC) composition by SDS-PAGE resolving the MyHC bands. Isoforms are identified with specific antibody against each of the types. The final data is calculated as sFo, i.e. maximum isometric force/CSA (mN/mm2), and compared across the 4 groups (placebo vs. Tofacitinib; 1h vs. 7 h) on each fiber type (Type I, II).

All investigators, including those carrying out assays, are blinded to treatment group (Tofacitinib vs. placebo). The patients are also blinded to group.

Statistical Analysis

The Tofacitinib and placebo group diaphragm and serratus anterior samples will be compared with respect to maximal force production and all biochemical and molecular markers measured at 1 h and 7 h (blood mononuclear cell lysates will be similarly compared with respect to JAK/STAT activation/phosphorylation). The more biologically relevant pre-post fold-change in these values will be tested as well as the log scale. Repeated measures two-way ANOVA, a mixed-effects model, paired t-test, or the non-parametric Wilcoxon signed rank test will be used to explore the differences between groups, as appropriate.

The intention of having the non-respiratory serratus anterior muscle as a control is to allow us to tease out the potential systemic effects of our model – effects that may potentially reduce the force generating capacity of all muscles -- from the effects of MV alone, which prior work suggests should be specific to respiratory muscles (e.g., the diaphragm). Our hope is that there will be little force reduction in the serratus fibers and more impressive force reduction in the diaphragm fibers. However, we will factor in changes in the contractility of the serratus muscle in at least two ways. We will initially analyze results in each muscle (serratus anterior and diaphragm) separately, using repeated two-way ANOVA. The two factors in the repeated two-way ANOVA will be ‘Treatment’ (2 groups: “Tofacitinib” and “Placebo”) and “Time” (measurements at 1 h and 7 h). This analysis will allow us to first determine whether any change in force (the dependent variable) is the result of the interaction between ‘Treatment” and “Time” in each muscle. We can then conceptually “subtract out” the loss of force incurred by the serratus fibers from the loss of force incurred by the diaphragm fibers, to get a somewhat qualitative idea of the impact of MV and the experimental treatment on respiratory vs. non-respiratory muscles. Only if this ‘corrected” force differs significantly in the Tofacitinib vs Placebo groups will be able to entirely reject the null hypothesis.

An additional way to analyze the impact of muscle type (serratus vs. diaphragm) that we will use is a mixed-effects model using 3 factors: ‘Treatment”, “Time”, and “Muscle Type”. The main effects of each of the factors -- Treatment, Time and Muscle Type -- and the effects of their interactions, can thereby be estimated after adjusting by other covariates. Three two-factor interactions will estimate three difference-in-difference (DID) effects in this analysis.

To address the possibility of results being confounded by malignancy, we will also study those patients who undergo esophagectomy for non-malignant diagnoses (e.g., late achalasia, benign esophageal stricture). We perform approximately 10 of these operations yearly, such that in two years we would likely have sufficient numbers of benign patients to compare their results with reasonable power to those in patients harboring malignancy. All statistical analyses will be performed using SAS software version 9.4 (SAS Institute Inc. NC, USA).

Population Size/Power analysis

The n=56 subjects planned for this randomized study is derived from a sample size estimation using our preliminary human molecular data similar to that shown in Figure 2. In that cohort, the mean measured pSTAT3/STAT3 at 1 h MV was ~1.0 with a standard deviation of ~0.5, and we saw an ~ 50% increase in mean pSTAT3/STAT3 level at the 7 h time point with a standard deviation of ~0.5. Choosing a 60% reduction by Tofacitinib in the degree of marker elevation at 7 h MV (chosen to likely represent a sufficient therapeutic impact to support proceeding to ICU-based clinical trials), assuming a conservative correlation of 0.3 between the repeated measures, and requiring a minimum power 80% for the paired design, the calculated, desired sample size was 56, with 28 subjects in each group.

Results

The study is ongoing, having begun to accrue subjects in September 2019. Results are forthcoming.

Discussion and Conclusions:

The trial we describe here represents the first, randomized, placebo-controlled, mechanistic clinical trial of a drug to prevent ventilator-induced diaphragm dysfunction. VIDD is a common and serious problem in our ICUs, responsible for many cases of ventilator-dependence, resulting in major morbidity and mortality and substantial costs to our healthcare system. A drug that could prevent VIDD by early delivery to patients who require MV would very likely have an important clinical impact, reducing duration of MV, downstream complications of MV, and ICU morbidity and mortality. The recently begun clinical trial described here builds upon recent basic discoveries by our group and others that identified the centrality of the JAK/STAT pathway to the development of VIDD. The successful completion of the proposed study will move this discovery toward a full-scale, ICU-based clinical trial, with clinical outcome variables, by providing evidence that JAK inhibition in fact prevents human ventilator-induced diaphragm weakness.

Highlights.

• Diaphragm weakness prolongs MV in many ventilated patients, impacting outcomes.

• There is no current drug therapy for this important problem.

• Inhibition of the JAK/STAT pathway prevents diaphragm weakness in ventilated rats.

• We describe an ongoing clinical trial of JAK inhibition (Tofacitinib) in ventilated patients.

• Outcome measures will be single fiber diaphragm force studies and molecular.