Abstract
BACKGROUND.
Cystic fibrosis (CF) pulmonary exacerbation (PEx) treatment guidelines suggest that Pseudomonas aeruginosa (Pa) airway infection be treated with two antipseudomonal agents.
METHODS.
We retrospectively studied treatment responses for STOP2 PEx treatment trial (NCT02781610) participants with a history of Pa infection. Mean lung function and symptom changes from intravenous (IV) antimicrobial treatment start to Visit 2 (7 to 10 days later) were compared between those receiving one, two, and three+ antipseudomonal classes before Visit 2 by ANCOVA. Odds of PEx retreatment with IV antimicrobials within 30 days and future IV-treated PEx hazard were modeled by logistic and Cox proportional hazards regression, respectively. Sensitivity analyses limited to the most common one-, two-, and three-class regimens, to only IV/oral antipseudomonal treatments, and with more stringent Pa infection definitions were conducted.
RESULTS.
Among 751 participants, 50 (6.7%) were treated with one antipseudomonal class before Visit 2, while 552 (73.5%) and 149 (19.8%) were treated with two and with three+ classes, respectively. Females and participants with a negative Pa culture in the prior month were more likely to be treated with a single class. The most common single, double, and triple class regimens were beta-lactam (BL; n=42), BL/aminoglycoside (AG; n=459), and BL/AG/fluoroquinolone (FQ; n=73). No lung function or symptom response, odds of retreatment, or future PEx hazard differences were observed by number of antipseudomonal classes administered in primary or sensitivity analyses.
CONCLUSIONS.
We were unable to identify additional benefit when multiple antipseudomonal classes are used to treat PEx in people with CF and Pa.
Keywords: Pulmonary exacerbation, P. aeruginosa, antipseudomonal classes
BACKGROUND
Chronic Pseudomonas aeruginosa in cystic fibrosis (CF) airways has been associated with accelerated lung disease progression [1,2] and increased mortality risk [3,4]. Accordingly, P. aeruginosa garners particular attention when people with CF experience pulmonary exacerbations (PEx), acute periods of increased respiratory signs and symptoms often accompanied by a drop in lung function [5]. For decades, CF PEx treatment guidelines have suggested that two antipseudomonal antimicrobials be included when treating PEx [6], although objective evidence of the clinical benefit of this practice has been lacking [7-9]. Early treatment guidelines didn’t specify that the two antipseudomonal agents be from different mechanistic classes (i.e., aminoglycosides, beta-lactams, fluoroquinolones, or polymyxins), but this was implied in a statement that “In most cases, one of these two will be an aminoglycoside” [6].
Use of two antipseudomonal agents has been rationalized for reducing emergence of acquired antimicrobial resistance and/or providing a potential for synergy of action [6], but the former has not been routinely observed among the few CF clinical trials comparing antipseudomonal monotherapies to combination therapies [10-13] and in vitro data suggest that the latter occurs in only a minority of instances [14]. On the contrary, some randomized trials have reported that combination antipseudomonal therapy was associated with increased antimicrobial resistance [11,12] and if/when synergy occurs, it has not been adequate to be associated with a detectable difference in mean clinical responses [10-13]. In fact, there are no data to suggest that choosing antimicrobial combinations based on in vitro synergy testing is associated with improved PEx treatment response [15], a main justification for reduced emphasis on in vitro antimicrobial synergy testing for choosing CF antimicrobials treatments [7,16]. The uncertain potential for greater benefit with combination antipseudomonal PEx treatment is difficult to balance against the increased costs, treatment complexity, and potential for systemic toxicity (including ototoxicity [17] and nephrotoxicity [18]) and selection for multiply resistant organisms associated with the practice [8].
The Standardized Treatment of Pulmonary Exacerbations (STOP) program was initiated by the US CF Foundation to design interventional trials to identify best practices for the management of PEx in CF. A recent multicenter trial (STOP2; NCT02781610) compared clinical responses to different intravenous (IV) antimicrobial PEx treatment durations [19]. In STOP2, investigators were provided a list of acceptable antimicrobial treatments and recommended dosages but given freedom to choose from that list based upon their knowledge of a participant’s airway microbiology and previous treatment responses. To maintain consistency in treatment regimens, the protocol required investigators to include two antipseudomonals when treating participants with P. aeruginosa, as in prior recommendations [6, 7]. Herein we describe antipseudomonal treatment choices made by investigators during STOP2 and compare clinical responses across numbers and types of antipseudomonal classes included in PEx treatments.
METHODS
The STOP2 study primary results have been described previously [19]. Briefly, STOP2 was a multi-center, randomized, controlled, clinical trial in PEx among adults with CF. After 7 to 10 days of IV antimicrobial treatment (at study Visit 2), participants exhibiting pre-specified lung function and symptom improvements were randomized to receive 10 or 14 total days of antimicrobial treatment; all others were randomized to receive 14 or 21 total days of treatment. Treatment responses were evaluated 14 days after the end of antimicrobial treatments (at Visit 3), ranging from 24 to 35 days after treatment start. Among CF adults with early PEx treatment improvement, mean ppFEV1 change from admission with 10 days of IV antimicrobial treatment was not inferior to mean change with 14 days of treatment. For participants with less improvement after one week, mean ppFEV1 change with 21 days of treatment was not superior to mean change with 14 days of treatment [19].
For the current retrospective analyses, only those STOP2 participants reported as having P. aeruginosa as an infecting airway opportunist within 30 days of screening or who had record of a positive P. aeruginosa culture in the previous year in the CF Foundation Patient Registry (CFFPR) were included. For each participant, antipseudomonal agents administered during STOP2 were categorized into classes irrespective of delivery route as aminoglycoside (AG), beta-lactam (BL), fluoroquinolone (FQ), or polymyxin (PX) (Supplemental Table S1). Participants were grouped by the number and combinations of antipseudomonal classes administered prior to study Visit 2 (Day 7 to 10), as well as by whether additional antipseudomonal agents were prescribed on or after Visit 2.
When available, participant mean ppFEV1 during the prior six months, IV antimicrobial treated PEx in the prior year, and IV antimicrobial treatment for PEx after Visit 3 were collected from the CFFPR. Mean ppFEV1 changes between visits were adjusted for age, mean ppFEV1 in the prior six months, ppFEV1 drop from prior average to Visit 1, and ever vs. never hospitalized during STOP2 (covariates selected a priori). Mean Chronic Respiratory Infection Symptom Score (CRISS) changes between visits were adjusted for the same covariates plus Visit 1 CRISS score. Higher CRISS scores indicate increased respiratory symptoms.
Demographic characteristics associated with treatment with a single antipseudomonal class prior to Visit 2 were assessed by logistic regression. Mean ppFEV1 and CRISS changes from Visit 1 to Visit 2 were compared between participants treated with one, two, or three or more antipseudomonal classes prior to Visit 2 by ANCOVA. Demographic, treatment, and treatment response characteristics at Visit 2 associated with addition of any new additional antipseudomonal agent on or after Visit 2 were assessed by logistic regression using a parsimonious approach to adjustment for confounders. Mean ppFEV1 and CRISS changes at Visit 2 and from Visit 2 to Visit 3 were compared between those receiving versus not receiving a new antipseudomonal agent on or after Visit 2 by ANCOVA. Odds of IV antimicrobial retreatment for PEx within 30 days after treatment cessation were compared across antipseudomonal class treatment subgroups by logistic regression and future IV-treated PEx hazards were compared by proportional hazards modeling, both adjusted for age, sex, mean ppFEV1 in the prior six months, ppFEV1 drop from prior average to Visit 1, any hospitalization during STOP2, any CFTR modulator use, and prior–year PEx number (≤1 vs >1), selected a priori.
As sensitivity analyses, comparisons were repeated using more stringent P. aeruginosa infection definitions which limited analyses to those participants a) with P. aeruginosa isolation within 30 days of Visit 1, b) with >1 P. aeruginosa isolation in the prior year, and c) with P. aeruginosa isolation in each of 2 years prior to STOP2. In addition, comparisons were repeated for the subset of participants receiving only the most frequent antimicrobial regimens associated with one, two, and three or more antipseudomonal classes administered prior to Visit 2, as well as repeated using only IV/oral treatments and excluding all inhaled antipseudomonal treatments from consideration. For each analysis, only those with data available at admission and Visits 2 and/or 3 were included, without imputation. Statistical analyses were performed using MedCalc® Statistical Software version 20.011 (MedCalc Software Ltd, Ostend, Belgium; https://www.medcalc.org; 2021) without adjustment for multiplicity of testing.
RESULTS
In all, 751 STOP2 participants ≥18 years old were included in analyses. Prior to Visit 2, 50 participants (6.7%) were treated with a single antipseudomonal class; 552 participants (73.5%) and 149 participants (19.8%) were treated with two and with three or more classes, respectively (Table 1). CFTR modulator use among 278 participants (37.0%; Table 2) did not differ across antipseudomonal class regimens (Supplemental Table S3). Only mean Visit 1 CRISS scores differed between participants using versus not using modulators (Supplemental Table S2). The most common antipseudomonal regimen prior to Visit 2 was the combination of AG/BL (n=459; 61.1%). On or after Visit 2, 101 participants (13.4%) received at least one additional (new) antipseudomonal agent, of which less than half (42, 5.6%) were of a new antipseudomonal class.
Table 1.
Participants administered antipseudomonal agents by class and regimen prior to Visit 2
| Antipseudomonal | Antipseudomonal Agents Administered Prior to Visit 2 | Total Participants (N=751) |
Antipseudomonal Classes |
|||||
|---|---|---|---|---|---|---|---|---|
| Class Regimen |
One (n=32) |
Two (n=439) |
Three (n=219) |
Four (n=45) |
Five (n=14) |
Six (n=2) |
||
| AG | 6 | 0 | 0 | 0 | 0 | 0 | 6 | One (n=50) |
| BL | 24 | 14 | 4 | 0 | 0 | 0 | 42 | |
| FQ | 1 | 0 | 0 | 0 | 0 | 0 | 1 | |
| PX | 1 | 0 | 0 | 0 | 0 | 0 | 1 | |
| AG/BL | 0 | 357 | 92 | 9 | 1 | 0 | 459 | Two (n=552) |
| AG/FQ | 0 | 2 | 0 | 0 | 0 | 0 | 2 | |
| AG/PX | 0 | 1 | 0 | 0 | 0 | 0 | 1 | |
| BL/FQ | 0 | 26 | 14 | 0 | 0 | 0 | 40 | |
| BL/PX | 0 | 36 | 11 | 0 | 0 | 0 | 47 | |
| PX/FQ | 0 | 3 | 0 | 0 | 0 | 0 | 3 | |
| AG/BL/FQ | 0 | 0 | 54 | 12 | 5 | 2 | 73 | Three or more (n=149) |
| AG/BL/PX | 0 | 0 | 38 | 15 | 3 | 0 | 56 | |
| AG/FQ/PX | 0 | 0 | 1 | 1 | 0 | 0 | 2 | |
| BL/FQ/PX | 0 | 0 | 5 | 2 | 2 | 0 | 9 | |
| AG/BL/FQ/PX | 0 | 0 | 0 | 6 | 3 | 0 | 9 | |
AG, aminoglycoside; BL, beta-lactam; FQ, fluoroquinolone; PX, polymyxin
Table 2. Participant demographics stratified by number of antipseudomonal classes administered before Visit 2.
Percentages are for columns unless otherwise noted
| Antipseudomonal Classes Administered before Visit 2 | ||||
|---|---|---|---|---|
| Demographics at Visit 1 | One | Two | Three or more | P-a |
| n, (% of 751 Pa–positive participants) | 50 (6.7%) | 552 (73.5%) | 149 (19.8%) | |
| Female sex, n (%) | 33 (66.0%) | 277 (50.2%) | 69 (53.7%) | .053 |
| Mean age, years [95% CI] | 32.8 [29.5, 36.0] | 30.5 [29.7, 31.3] | 31.8 [30.2, 33.3] | .140 |
| Mean prior 6-month ppFEV1 average [95% CI]b,c | 52.2 [46.1, 58.4] | 51.9 [50.3, 53.6] | 53.8 [50.6, 57.0] | .609 |
| Prior 6-month ppFEV1 group, n (%) b | ||||
| <40 | 17 (34.0%) | 173 (31.3%) | 35 (23.5%) | .567 |
| 40 to <70 | 21 (42.0%) | 257 (46.6%) | 79 (53.0%) | |
| ≥70 | 11 (22.0%) | 104 (18.8%) | 29 (19.5%) | |
| Unknown | 1 (2.0%) | 18 (3.3%) | 6 (4.0%) | |
| Mean Visit 1 ppFEV1 [95% CI] | 48.4 [42.4, 54.3] | 47.9 [46.2, 49.5] | 48.4 [45.6, 51.3] | .946 |
| Mean ppFEV1 drop from 6-month average [95% CI]c | 4.1 [2.5, 5.8] | 3.8 [3.1, 4.5] | 5.1 [3.7, 6.5] | .233 |
| ppFEV1 drop group, n (%) | ||||
| >10 | 8 (16.0%) | 86 (15.6%) | 28 (18.8%) | .933 |
| >5 to 10 | 7 (14.0%) | 97 (17.6%) | 29 (19.5%) | |
| >0 to 5 | 23 (46.0%) | 231 (41.8%) | 59 (39.6%) | |
| ≤0 (increase) | 11 (22.0%) | 120 (21.7%) | 27 (18.1%) | |
| Unknown | 1 (2.0%) | 18 (3.3%) | 6 (4.0%) | |
| Mean Visit 1 CRISS [95% CI] | 51.5 [48.4, 54.6] | 50.5 [49.6, 51.4] | 52.7 [51.0, 54.5] | .077 |
| Two or more PEx in the prior year, n (%) | 31 (62.0%) | 325 (58.9%) | 92 (61.7%) | .524 |
| MRSA positive in previous 12 months, n (%)b | 16 (32.0%) | 188 (34.1%) | 41 (27.5%) | .318 |
| NTM positive in prior 12 months, n (%)b | 9 (18.0%) | 60 (10.9%) | 18 (12.1%) | .313 |
| Receiving CFTR modulators at Visit 1 | 15 (30.0%) | 208 (37.7%) | 55 (36.9%) | .560 |
Chi-square tests for proportions and ANOVA for means.
From Cystic Fibrosis Foundation Patient Registry.
N=49 for One Antipseudomonal Class, N=534 for Two Antipseudomonal Classes, and N=143 for Three+ Antipseudomonal Classes.
ppFEV1, percent predicted forced expiratory volume in 1 sec; MRSA, methicillin-resistant Staphylococcus aureus; NTM, non-tuberculous mycobacteria
No significant demographic differences were observed between individuals administered one, two, or three or more antipseudomonal classes prior to Visit 2 (Table 2). However, the odds of being treated with a single class were significantly greater for females and those participants with P. aeruginosa-negative culture results within 30 days of Visit 1; odds were significantly lesser for participants presenting with new or increased adventitial sounds on chest exam and with increased congestion or changes in sputum (Supplemental Figure S1).
Adjusted mean ppFEV1 and CRISS changes from Visit 1 to Visit 2 did not differ by the number of antipseudomonal classes administered (Figure 1, Table 3). However, mean changes from Visit 1 to Visit 2 were significantly worse for the 13.4% of participants treated with a new antipseudomonal agent on or after Visit 2 than for those not treated with a new agent (mean ppFEV1 change difference of −2.4 [95% CI −4.0, −0.7], P=.0045; mean CRISS change difference of 2.9 [0.5, 5.3], P=.0164). This was also true for the 5.6% of participants treated with a new antipseudomonal class on or after Visit 2 (mean ppFEV1 change difference of −2.8 [95% CI −5.2, −0.4], P = .024; mean CRISS change difference of 4.2 [0.4, 8.0], P = .0303). Odds of treatment with a new antipseudomonal agent on or after Visit 2 were associated with a lesser ppFEV1 response at Visit 2 and trended towards an association with a lesser CRISS response at Visit 2 (Figure 2). Number of antipseudomonal classes administered prior to Visit 2, age, sex, and baseline lung function were not associated with the decision to treat with a new antipseudomonal agent on or after Visit 2. In addition to poorer ppFEV1 response at Visit 2, having been treated with a single antipseudomonal class prior to Visit 2 was associated with treatment with a new antipseudomonal class on or after Visit 2, (Figure 2).
Figure 1. Mean lung function and symptom scores at study visits by number of antipseudomonal classes administered prior to Visit 2.
Upper panel: mean ppFEV1 values. Lower panel: mean CRISS values. Participants treated with one, two, or three or more antipseudomonal classes are presented as black, gray, and white circles, respectively. Bars are 95% confidence intervals (CI).
Table 3.
Mean ppFEV1 and CRISS changes by antipseudomonal classes administered prior to Visit 2.
| Antipseudomonal Classes Administered prior to Visit 2 | |||||||
|---|---|---|---|---|---|---|---|
| One | Two | Three or more | |||||
| (n) | Mean [95% CI] | (n) | Mean [95% CI] | (n) | Mean [95% CI] | P a | |
| ppFEV1 change, Visit 1 to Visit 2 | 49 | 6.6 [4.5, 8.8] | 534 | 7.0 [6.3, 7.6] | 143 | 6.2 [4.9, 7.4] | .54 |
| Difference from One Antipseudomonal Class | - | 0.3 [−2.5, 3.5] | −0.5 [−3.5, 2.6] | 1.0, 1.0 | |||
| ppFEV1 change, Visit 2 to Visit 3 | 47 | −0.0 [−2,1, 2.2] | 505 | −0.3 [−1.0, 0.3] | 129 | 0.2 [−1.1, 1.5] | .75 |
| No new antipseudomonal agent on or after Visit 2 | 38 | −0.1 [−2.6, 2.3] | 437 | −0.5 [−1.2, 0.2] | 109 | −0.2 [−1.6, 1.3] | .89 |
| New antipseudomonal agent on or after Visit 2 | 9 | 0.6 [−4.4, 5.5] | 68 | 0.8 [−1.0, 2.6] | 20 | 2.2 [−1.2, 5.5] | .77 |
| No new antipseudomonal class on or after Visit 2 | 40 | 0.1 [−2.3, 2.5] | 477 | −0.4 [−1.1, 0.3] | 122 | 0.2 [−1.1, 1.6] | .71 |
| New antipseudomonal class on or after Visit 2 | 7 | −0.5 [−5.0, 4.0] | 28 | 0.3 [−1.8, 2.5] | 7 | 0.2 [−4.3, 4.7] | .94 |
| CRISS change, Visit 1 to Visit 2 | 49 | −13.9 [−17.0, −10.7] | 534 | −17.3 [−18.3, −16.4] | 143 | −15.9 [−17.8, −14.1] | .073 |
| Difference from One Antipseudomonal Class | - | −3.4 [−7.5, 0.6] | −2.1 [−6.5, 2.4] | .12, .80 | |||
| CRISS change, Visit 2 to Visit 3 | 47 | −5.3 [−8.9 −1.7] | 506 | −0.6 [−1.7, 0.5] | 130 | −0.8 [−2.9, 1.4] | .043 |
| No new antipseudomonal agent on or after Visit 2 | 38 | −4.8 [−8.6, −1.0] | 438 | −0.3 [−1.4, 0.8] | 110 | −0.2 [−2.5, 2.1] | .084 |
| New antipseudomonal agent on or after Visit 2 | 9 | −7.3 [−16.8, 2.2] | 68 | −2.0 [−5.5, 1.4] | 20 | −4.9 [−11.6, 1.8] | .50 |
| No new antipseudomonal class on or after Visit 2 | 40 | −4.7 [−8.5, −1.0] | 478 | −0.6 [−1.7, 0.5] | 123 | −0.3 [−2.5, 1.8] | .11 |
| New antipseudomonal class on or after Visit 2 | 7 | −10.2 [−22.4, 1.9] | 28 | −0.7 [−6.6, 5.1] | 7 | −7.8 [−20.8, 5.1] | .26 |
ANCOVA for differences between numbers of classes administered adjusted for adjusted for age, average prior 6-month ppFEV1, ppFEV1 drop from 6-month average at Visit 1, ever/never hospitalized during STOP2, and CRISS score at Visit 1 (CRISS means only). Bolding denotes 95% confidence intervals (CI) not including zero or P-values <.05.
Figure 2. Odds ratios for treatment with a new antipseudomonal agent or class on or after Visit 2.
Black circles, odds ratios for treatment with a new antipseudomonal agent on or after Visit 2. Gray circles, odds ratios for treatment with a new antipseudomonal class on or after Visit 2. Bars are 90% confidence intervals (CI). CFFPR, CF Foundation Patient Registry.
Mean ppFEV1 and CRISS changes from Visit 2 to Visit 3 were nominal for all participants, with ppFEV1 decreasing 0.2 [95% CI −0.8, 0.4] and CRISS decreasing 0.9 [−0.0, 1.9]. Mean changes did not differ for participants receiving versus not receiving a new agent on or after Visit 2 (ppFEV1 change difference 1.4 [−0.2, 3.1], P = .083; CRISS change difference −2.3 [−5.0, 0.4]; P = .095). Similar results were observed for subgroups of participants receiving one, two, and three or more of antipseudomonal classes administered prior to Visit 2 (Table 3). In contrast, mean CRISS changes from Visit 2 to Visit 3 differed by number of antipseudomonal classes administered prior to Visit 2 (P=.043), being significantly different than zero among those treated with a single antipseudomonal (mean CRISS change of −5.3 [−8.9, −1.7]) but not for those treated with more classes before Visit 2 (Table 3). Mean CRISS changes from Visit 2 to Visit 3 did not differ for those treated versus not treated with a new antipseudomonal agent on or after Visit 2 (CRISS change difference −2.3 [−5.0, 0.4], P=.095) and there were no differences in mean CRISS change between those treated versus not treated with a new agent across antipseudomonal treatment regimens (one, two, or three+ classes; Table 3).
Odds of retreatment with IV antimicrobials for PEx within 30 days of treatment cessation (an indication of treatment failure) and hazard for future PEx did not differ as a function of the number of antipseudomonal classes administered prior to Visit 2 (Figure 3). Older age, greater baseline ppFEV1, and any CFTR modulator use were associated with reduced future PEx hazard, while female sex and >1 IV-treated PEx in the prior year were associated with increased future PEx hazard (Figure 3).
Figure 3. Odds ratio for retreatment for PEx within 30 days of treatment end and hazard ratios for future PEx treated with IV antimicrobials.
Left panel, odds of retreatment. Right panel, future PEx hazard. Bars represent 95% confidence intervals (CI). AntiPa, antipseudomonal; CFFPR, CF Foundation Patient Registry.
No differences in treatment responses were observed across antipseudomonal class groups when analyses were limited to those participants with additional evidence of P. aeruginosa infection, including culture isolation within 30 days of Visit 1, >1 isolation in the prior year, or isolation in each of 2 prior years (Supplemental Tables S4 and S5). Participants receiving the most common antipseudomonal regimens containing one, two, or three or more classes (BL [n=42], AG/BL [n=459], and AG/BL/FQ [n=73]) did not experience different mean treatment responses, (Supplemental Table S6), nor did treatment groups defined based only on IV and oral (PO) antipseudomonal treatments (Supplemental Table S8).
DISCUSSION
This retrospective analysis of STOP2 trial results provided no evidence of differences in PEx treatment responses among those with P. aeruginosa airways infection as a function of the number of antipseudomonal classes administered. Current CF guidelines recommend treating these patients with two antipseudomonal agents [6] but acknowledge that there are few objective data to support this practice [7], a position supported by recent Cochrane Database reviews [8,9]. In fact, the handful of prospective randomized trials that have been conducted, most of which were relatively small, were unable to identify an immediate clinical benefit with respect to either lung function or symptom score associated with combination antipseudomonal PEx treatment [10-13]. One previous prospective study, an investigator-blinded, multicenter comparison of PEx treatment with azlocillin (a beta-lactam) alone versus tobramycin (an aminoglycoside) plus azlocillin, showed no difference in lung function or symptom score responses at treatment end but did report a difference in median time to next PEx in favor of combination treatment [12]. Unfortunately, that study did not stratify by prior-year PEx number at randomization and the associated time-to-next PEx analysis did not include adjustment for covariates known to be associated with future PEx hazard, including age, sex, lung disease stage, and number of prior-year PEx [20]. After adjusting for these known risk factors for future PEx treatment hazard in the current analyses, we did not observe a difference in future PEx hazard associated with number of antipseudomonal classes administered in either base or sensitivity analyses.
Interestingly, although modulator use within the STOP2 cohort was observed to be a significant covariate for future PEx hazard, it was not a significant covariate in ANCOVA adjustments for ppFEV1 or CRISS responses from Visit 1 (data not shown), suggesting that individuals prescribed CFTR modulators may be prone to less frequent, but similar, PEx presentations and responses as those not receiving modulators. This observation is consistent with a previous analysis of PEx treatment responses among study subjects receiving ivacaftor [21].
Important limitations of this analysis include its retrospective nature and our inability to account for known and unknown confounders in the way that prospective randomized studies can. Known confounders include, but are not limited to, lung disease stage, prior year PEx number, and sex. Unknown confounders might include relative airway P. aeruginosa density, interactions with viruses and other bacterial species, and presence of CF modifier genes. In addition, because CF “standard of care” is to treat P. aeruginosa PEx with multiple antipseudomonal classes, this retrospective analysis is unbalanced with respect to antipseudomonal regimens employed, making the identification of subtle differences in treatment groups and outcomes more difficult. Our base analysis included STOP2 participants who had been reported to have a positive P. aeruginosa culture within 30 days of Visit 1 or in the prior year based on CFFPR records, a singular characterization of the nature of P. aeruginosa infection. Conclusions drawn from sensitivity analyses with more stringent P. aeruginosa culture history definitions were no different, but these necessarily involved smaller sample sizes and thus had less power to identify subtle differences in response. Ours is not the first retrospective analysis that has been unable to identify a benefit of treating PEx in people with CF and P. aeruginosa with more than one antipseudomonal agent [22], but all such analyses are limited by the challenge of controlling for indication bias. Because antimicrobial treatment choices were not randomly assigned in STOP2 (e.g., women were more likely than men to be treated with a single antipseudomonal class, as were participants who did not culture P. aeruginosa within 30 days of Visit 1 despite history of positive culture), and because reasons for treatment choices were not collected, association should not be confused with causality. An additional limitation of this analysis is that it cannot account for the effects of different antipseudomonal treatment regimens on repeated PEx treatments over years, which people with CF experience. Our analyses focused on the relationships between antipseudomonal classes administered and treatment responses at Visit 2 to avoid confounding by addition of new agents that occurred on or after Visit 2. Addition of agents on or after Visit 2 in a minority of subjects was apparently driven by poorer than expected treatment responses but was no more likely to occur after treatment with a single antipseudomonal class than with multiple classes and there was no indication that these additions affected Visit 3 responses. In our primary analyses, we grouped all routes of antipseudomonal treatment together when defining the number of antipseudomonal classes in a regimen. This approach does not account for known differences in airway pharmacokinetics between inhaled and systemic treatments, the former generating extremely high airway drug concentrations but limited to ventilated areas and the latter with more modest concentrations but with access to obstructed areas of the lung. However, sensitivity analyses excluding inhaled treatments yielded the same result: no distinctions in response were observed as a function of the number of IV/PO antipseudomonal classes administered prior to Visit 2. Although four antipseudomonal classes (as defined by mechanism of action) were used in STOP2, the spectra of antibacterial activities across these classes are dissimilar and the breadths of activity within these classes are inconsistent, raising a question as to whether “counting classes” is the best categorization method. In particular, there are beta-lactam class members that are “narrower” in their antibacterial spectra (e.g., aztreonam, cefepime, and ceftazidime) and others that are substantially “broader” (e.g., piperacillin/tazobactam and meropenem), and it may be that these distinctions are relevant to treatment response when combined with other classes. Future analyses are planned to better describe possible relationships between antimicrobial spectrum and STOP2 treatment response. We did not impute demographic data missing from the CFFPR or follow-up ppFEV1 or CRISS data missing from STOP2 and do not know whether this missingness was at random or relative to antipseudomonal treatment decisions; however prior analyses of primary outcomes using all the STOP2 data supported a random missingness mechanism [19]. Fortunately, numbers of missing data were quite small and thus unlikely to have materially affected results. Finally, a primary justification for combining antipseudomonal classes, reduction of antimicrobial resistance emergence, was not tested in this analysis. However, a lack of supportive evidence from past clinical trials in which resistance emergence was studied combined with the knowledge that P. aeruginosa is rarely if ever eradicated during PEx treatment (and thus survives whatever treatment has been administered) raise questions with respect to this theoretical benefit.
Re-evaluation of the risk/benefit associated with combined aminoglycoside/beta-lactam treatment of P. aeruginosa has precedent. A large meta-analysis of randomized trials of sepsis treatment in immunocompromised patients concluded that combining aminoglycosides with beta-lactams provided no differences with respect to either P. aeruginosa resistance emergence or patient survival relative to use of beta-lactams alone but was associated with an increased risk of adverse events, including nephrotoxicity [23]. Given the potential for increased cost, complexity, and risk of harm associated with treating CF PEx with more than one antipseudomonal class, one would like to have objective evidence of a clear benefit to justify this practice, something which might only be achieved by a randomized prospective trial [9]. We found that the most frequent choice of a single antipseudomonal class during STOP2 was beta-lactam and the most common class combination was aminoglycoside and beta-lactam, suggesting that the most feasible prospective study might be one in which people with CF and P. aeruginosa who are to be treated for PEx are randomized 1:1 to be treated for a fixed duration with a beta-lactam either with or without an aminoglycoside. Given that aminoglycoside use has been shown to be associated with cochleotoxicity and vestibular toxicity that is irreversible [24], it would be reasonable to expect addition of an aminoglycoside in this context to provide a superior clinical response to treatment with a beta-lactam alone to justify its use. Our results suggest that such a trial would require >300 participants per arm to afford 90% power to detect at least a 2.5 ppFEV1 treatment associated response difference.
Supplementary Material
Financial Support:
The STOP2 study was supported by grants from the Cystic Fibrosis Foundation: FLUME15A0, FLUME17AB0, HELTSH15A0, GOSS15A0, SANDER14A0, SANDER17AB0, WEST15A0, WEST17AB0, WEST17Y5. PF is supported, in part, by the National Center for Advancing Translational Sciences of the National Institutes of Health under Grant Number UL1 TR001450. CG and SH are supported, in part, by the National Institutes of Health under Grant Number P30 DK089507
Footnotes
Conflicts of interest
The authors claim no financial conflicts of interest related to this work.
REFERENCES
- [1].Konstan MW, Morgan WJ, Butler SM, Pasta DJ, Craib ML, Silva SJ, Stokes DC, Wohl ME, Wagener JS, Regelmann WE, Johnson CA; Scientific Advisory Group and the Investigators and Coordinators of the Epidemiologic Study of Cystic Fibrosis. Risk factors for rate of decline in forced expiratory volume in one second in children and adolescents with cystic fibrosis. J Pediatr. 2007. Aug;151(2):134–9, 139.e1. [DOI] [PubMed] [Google Scholar]
- [2].Konstan MW, Wagener JS, VanDevanter DR, Pasta DJ, Yegin A, Rasouliyan L, Morgan WJ. Risk factors for rate of decline in FEV1 in adults with cystic fibrosis. J Cyst Fibros. 2012. Sep;11(5):405–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Liou TG, Adler FR, Fitzsimmons SC, Cahill BC, Hibbs JR, Marshall BC. Predictive 5-year survivorship model of cystic fibrosis. Am J Epidemiol 2001; 153(4): 345–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Mayer-Hamblett N, Rosenfeld M, Emerson J, Goss CH, Aitken ML. Developing cystic fibrosis lung transplant referral criteria using predictors of 2-year mortality. Am J Respir Crit Care Med. 2002; 166(12 Pt 1):1550–5. [DOI] [PubMed] [Google Scholar]
- [5].Ratjen F, Döring G. Cystic fibrosis. Lancet. 2003. Feb 22;361(9358):681–9. Review. [DOI] [PubMed] [Google Scholar]
- [6].Treatment of pulmonary exacerbation of cystic fibrosis. Clinical Practice Guidelines for Cystic Fibrosis. Bethesda, MD: Cystic Fibrosis Foundation; 1997. [Google Scholar]
- [7].Flume PA, Mogayzel PJ Jr, Robinson KA, Goss CH, Rosenblatt RL, Kuhn RJ, Marshall BC; Clinical Practice Guidelines for Pulmonary Therapies Committee. Cystic fibrosis pulmonary guidelines: treatment of pulmonary exacerbations. Am J Respir Crit Care Med 2009; 180(9): 802–8. [DOI] [PubMed] [Google Scholar]
- [8].Elphick HE, Scott A. Single versus combination intravenous anti-pseudomonal antibiotic therapy for people with cystic fibrosis. Cochrane Database Syst Rev. 2016. Dec 1;12(12):CD002007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Holland P, Jahnke N. Single versus combination intravenous anti-pseudomonal antibiotic therapy for people with cystic fibrosis. Cochrane Database Syst Rev. 2021. Jun 23;6(6):CD002007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].McLaughlin FJ, Matthews WJ Jr, Strieder DJ, Sullivan B, Taneja A, Murphy P, Goldmann DA. Clinical and bacteriological responses to three antibiotic regimens for acute exacerbations of cystic fibrosis: ticarcillin-tobramycin, azlocillin-tobramycin, and azlocillin-placebo. J Infect Dis. 1983. Mar;147(3):559–67. [DOI] [PubMed] [Google Scholar]
- [11].Padoan R, Cambisano W, Costantini D, Crossignani RM, Danza ML, Trezzi G, Giunta A. Ceftazidime monotherapy vs. combined therapy in Pseudomonas pulmonary infections in cystic fibrosis. Pediatr Infect Dis J. 1987. Jul;6(7):648–53. [DOI] [PubMed] [Google Scholar]
- [12].Smith AL, Doershuk C, Goldmann D, Gore E, Hilman B, Marks M, Moss R, Ramsey B, Redding G, Rubio T, Williams-Warren J, Wilmott R, Wilson HD, Yogev R. Comparison of a beta-lactam alone versus beta-lactam and an aminoglycoside for pulmonary exacerbation in cystic fibrosis. J Pediatr. 1999. Apr;134(4):413–21. [DOI] [PubMed] [Google Scholar]
- [13].Master V, Roberts GW, Coulthard KP, Baghurst PA, Martin A, Roberts ME, Onishko CR, Martin AJ, Linke RJ, Holmes M, Jarvinen A, Kennedy D, Colebatch KA, Hansman D, Parsons DW. Efficacy of once-daily tobramycin monotherapy for acute pulmonary exacerbations of cystic fibrosis: a preliminary study. Pediatr Pulmonol. 2001. May;31(5):367–76. [DOI] [PubMed] [Google Scholar]
- [14].Balke B, Hogardt M, Schmoldt S, Hoy L, Weissbrodt H, Häussler S. Evaluation of the E test for the assessment of synergy of antibiotic combinations against multiresistant Pseudomonas aeruginosa isolates from cystic fibrosis patients. Eur J Clin Microbiol Infect Dis. 2006. Jan;25(1):25–30. [DOI] [PubMed] [Google Scholar]
- [15].Smith S, Ratjen F, Remmington T, Waters V. Combination antimicrobial susceptibility testing for acute exacerbations in chronic infection of Pseudomonas aeruginosa in cystic fibrosis. Cochrane Database Syst Rev. 2020. May 15;5(5):CD006961. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Aaron SD. Antibiotic synergy testing should not be routine for patients with cystic fibrosis who are infected with multiresistant bacterial organisms. Paediatr Respir Rev. 2007. Sep;8(3):256–61. [DOI] [PubMed] [Google Scholar]
- [17].Harruff EE, Kil J, Ortiz MGT, Dorgan D, Jain R, Poth EA, Fifer RC, Kim YJM, Shoup AG, Flume PA. Ototoxicity in cystic fibrosis patients receiving intravenous tobramycin for acute pulmonary exacerbation: Ototoxicity following tobramycin treatment. J Cyst Fibros. 2021. Mar;20(2):288–294. [DOI] [PubMed] [Google Scholar]
- [18]].LeCleir LK, Pettit RS. Piperacillin-tazobactam versus cefepime incidence of acute kidney injury in combination with vancomycin and tobramycin in pediatric cystic fibrosis patients. Pediatr Pulmonol. 2017. Aug;52(8):1000–1005. [DOI] [PubMed] [Google Scholar]
- [19].Goss CH, Heltshe SL, West NE, Skalland M, Sanders DB, Jain R, Barto TL, Fogarty B, Marshall BC, VanDevanter DR, Flume PA; STOP2 Investigators. A Randomized Trial of Antimicrobial Duration for Cystic Fibrosis Pulmonary Exacerbation Treatment. Am J Respir Crit Care Med. 2021. Sep 1. doi: 10.1164/rccm.202102-0461OC. Epub ahead of print. [DOI] [Google Scholar]
- [20].VanDevanter DR, Morris NJ, Konstan MW. IV-treated pulmonary exacerbations in the prior year: An important independent risk factor for future pulmonary exacerbation in cystic fibrosis. J Cyst Fibros. 2016. May;15(3):372–379. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Flume PA, Wainwright CE, Tullis DE, Rodriguez S, Niknian M, Mark M, Davies JC, Wagener JS. Recovery of lung function following a pulmonary exacerbation in patients with cystic fibrosis and the G551D-CFTR mutation treated with ivacaftor. J Cyst Fibros 2018; 17: 83–88. [DOI] [PubMed] [Google Scholar]
- [22].Cogen JD, Faino AV, Onchiri F, Hoffman LR, Kronman MP, Nichols DP, Rosenfeld M, Gibson RL. Association Between Number of Intravenous Antipseudomonal Antibiotics and Clinical Outcomes of Pediatric Cystic Fibrosis Pulmonary Exacerbations. Clin Infect Dis. 2021. Jun 8:ciab525. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23].Paul M, Benuri-Silbiger I, Soares-Weiser K, Leibovici L. Beta lactam monotherapy versus beta lactam-aminoglycoside combination therapy for sepsis in immunocompetent patients: systematic review and meta-analysis of randomised trials. BMJ. 2004. Mar 20;328(7441):668. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Jiang M, Karasawa T, Steyger PS. Aminoglycoside-Induced Cochleotoxicity: A Review. Front Cell Neurosci. 2017. Oct 9;11:308. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.