The current study, conducted at 18 centers in the United States from June 2010 to August 2010, was a randomized double-blind, double-dummy trial to determine the clinical efficacy, safety, and tolerability of JNJ-Q2 (250 mg, twice a day[BID]) versus linezolid (600 mg, BID) in patients with ABSSSI. The dose of JNJ-Q2 was chosen based upon pharmacokinetic/pharmacodynamic (PK/PD) modeling using the ratio for the area under the time-concentration curve (AUC) and the MIC as the marker of interest and targeting a 1-log kill of FQR-MRSA while limiting potential side effects. A time and events overview of the study design, showing the treatment and evaluation schedule, is provided in . The study was conducted in accordance with all relevant federal guidelines and institutional policies, including, but not limited to, informed patient consent prior to enrollment and institutional review.
Adult males and nonpregnant, nonlactating females, age 18 to 70 years inclusive, with an ABSSSI were stratified by infection type (wound infection, cellulitis, or severe abscess) and fever (≥38.0°C) and randomly assigned (1:1) to receive either JNJ-Q2 or linezolid. Investigators were instructed to stop medication when the infection was considered clinically cured after a minimum treatment period of 7 days and a maximum treatment period of 14 days. Specimens from the primary infection site were collected for culture. Confirmation testing was performed on bacterial isolates for identification and susceptibility at the central microbiology reference laboratory (JMI Laboratories, North Liberty, IA). Qualifying ABSSSI included any wound infection, deep cellulitis, or severe abscess with a minimum lesion size of 75 cm2
. Infections were not likely to be cured with surgical incision alone and did not require other antifungal, antimycobacterial, or other antibacterial agents in addition to study medication. Lesion surface area was determined by the product of the maximum length and width of the cross-sectional diameters of the leading edge of either the erythema or induration, whichever was greater. To be enrolled, patients were required to have at least 3 of the following 5 signs: purulent drainage/discharge; erythema with/without induration; fluctuation; heat/localized warmth; pain/tenderness. Additionally, at least one of the following six systemic findings had to be present: temperature of ≥38°C (100.4°F); hypothermia (<36.0°C [96.8°F]); white blood cell count of ≥10,000/μl; ≥15% band forms; leukopenia; lymphangitis/lymphadenopathy relative to the area of the infection (10
). Patients who received 1 dose of a systemic antibiotic (with a half-life of 12 h or less) 24 h prior to the first dose were permitted in the study. Patients with surgical incisional wound infections who received prophylactic antibiotics ≥48 h before enrollment were also permitted in the study. All patients with infections involving prosthetic materials or foreign bodies (except cellulitis and purulence at the insertion site of an intravenous catheter), coexisting conditions, such as decubitus ulcer, diabetic foot ulcer, septic arthritis, gangrene or gas gangrene, burns, current evidence of deep vein thrombosis, or a severely compromised immune system, were not eligible for enrollment.
Patients who needed any other systemic antibiotic treatment or antibiotic treatment beyond 14 days were considered clinically not cured; investigators could choose to withdraw patients from the study at any time and treat with alternative therapy if the patient was not improving regardless of the presence or absence of a Gram-negative organism. The follow-up period included a test-of-cure (TOC) visit 10 to 14 days after randomization and a short-term follow-up (SFU) visit that occurred 2 to 14 days after the last dose of study medication. A secondary or long-term follow-up visit occurred between 84 and 98 days postrandomization.
Assessments included signs and symptoms of infection and infection site measurements. Postrandomization patients were required to take their oral body temperature at least 5 times daily while awake as well as just before consumption of an antipyretic through the SFU visit. All patients were provided with the same standard digital thermometers and daily temperature cards for recording the actual temperature reading as well as the date and time when they took their temperature. Signs and symptoms (erythema, edema/induration, purulence [including drainage], tenderness/pain, and lymphangitis/lymphadenopathy) of the primary lesion were assessed on a scale of 0 to 3 (none, mild, moderate, or severe).
Safety was assessed based on adverse event (AE) monitoring, 12-lead electrocardiogram measurements, physical examination findings, vital sign measurements, visual acuity assessments, and clinical laboratory assessments.
A 70% cure rate for linezolid, a 75% cure rate for JNJ-Q2, and a noninferiority margin of 15% were assumed in determining study sample size. These assumptions determined that a target of 80 patients per treatment group (i.e., 160 patients in total) would yield over 80% power to determine that JNJ-Q2 is not inferior to linezolid, assuming a 1-sided α-level of 0.025. The primary efficacy analysis of the ITT set (see below) was conducted on the composite endpoint of (i) cessation of spread of the primary infection site lesion and (ii) stable temperature below 38°C, evaluated at 36 to 84 h after the first dose of study medication. Although the protocol stipulated 48 to 72 h for the evaluation, the prospective statistical analysis plan (SAP) allowed an additional 12 h on both sides of the time frame, broadening the window to 36 to 84 h to allow scheduling flexibility for the sites and patients, especially those who were randomized on the afternoon of day 1 and required a morning visit on day 3. Stable temperature was defined as successive temperature readings falling within 2 standard errors of measurement of one another by 72 h postrandomization. The standard error was determined by regression of temperature measurements over time for all patients irrespective of treatment assignment.
A primary responder was any subject who met success on both elements of the composite end point, i.e., lack of spread of lesion and afebrile, during the evaluation window. Any other result pattern, including patients outside the visit window, was assigned as a treatment failure. The primary efficacy analysis was performed using logistic regression (14
), where treatment and wound type were main effects and baseline temperature and lesion size were covariates.
analyses were performed on the ITT set by using the same lesion response criteria but also meeting the FDA definition of afebrile and using the 48- to 72-h window (10
). This analysis of early clinical response criteria was supplemented with a risk difference (RD) analysis (4
). provides a comparison of early end point criteria as defined by the protocol, the prospective SAP for the study, and FDA definitions.
Comparison of early responder analysis criteria
Other prespecified secondary analyses on the ITT set included clinical response rates as determined by the investigator at the later time points on day 8, TOC, and SFU. Missing clinical responses or patient visits occurring outside the prospectively defined visit window, regardless as to investigator assessment, were assigned as clinical failures in the statistical analyses. These later time point clinical responses were compared using logistic regression (14
), where treatment and wound type were main effects. Risk difference analyses (4
) were conducted in addition to logistic regression (14
) modeling for responses on day 8, TOC, and SFU.
Other post hoc
analyses were done using a composite end point of early assessment of lesion response and investigator assessment of clinical cure at the TOC time point. The first secondary end point noted in the draft FDA ABSSSI guidance includes “evaluation and objective measurements of the infection site and overall assessment as clinical response or failure” (10
). The ITT set was analyzed with respect to the number of patients who had both a clinical response of “cured” at the TOC visit and who were also considered lesion responders, i.e., those with cessation of spread or reduction in the size of the primary infection site lesion at visit 3 (day 4). This included only patients with a TOC visit within the appropriate analysis window of 10 to 14 days from the date of randomization. These analyses were conducted using both windows for lesion response, i.e., the window as defined in the SAP (36 to 84 h; primary analysis lesion responder) and the narrower FDA window (48 to 72 h; FDA lesion responder ).
The intent-to-treat analysis set included patients receiving at least one dose of drug and was employed for the primary efficacy analysis and other post hoc analyses. The microbiological ITT (mITT) set contained patients in the ITT set that had microbiologic confirmation of disease. Patients in the clinically evaluable (CE) set were mITT patients that also met inclusion/exclusion criteria, had adequate description of the infected area/lesion, had adequate follow-up (at or after the timing for the TOC visit), and a culture obtained (if an appropriate site to culture was available), received at least 80% of planned study medication therapy, had a TOC visit from 10 to 14 days postrandomization, and did not receive any concomitant systemic antibacterial therapy. The microbiologically evaluable (ME) set consisted of CE patients who had a pathogen isolated at baseline that was susceptible to study medications.
A total of 161 patients were randomized, with 83 patients receiving JNJ-Q2 and 78 patients receiving linezolid. Demographic characteristics were generally similar between treatment groups. The mean age was 36.9 (range, 18 to 69) years. The majority of patients were male (64.6% [104 of 161]), and most patients were white (79.5% [128 of 161]). Details about baseline diagnosis and patient disposition are shown in . Although patients with diabetic foot ulcers were not eligible for enrollment, patients with diabetes were permitted in the study and represented 13% of patients enrolled (16.9% in the JNJ-Q2 arm and 9.0% in the linezolid arm).
Subject disposition and diagnosis
Approximately one-third of patients in each treatment group had abscesses, wound infections, and cellulitis. Although the initial lesion sizes were large (mean and median lesion sizes were 266 and 162 cm2, respectively, per study sample), the overall rate of fever across both groups was low (4.3%). A total of 21 patients, 10 (12%) on JNJ-Q2 and 11(13.9%) on linezolid, received prior antibiotic therapy as allowed by the protocol. Discontinuations due to adverse events were low (2.4% for JNJ-Q2 and 2.6% for linezolid), and there was no use of rescue antibiotics.
Approximately 75% of patients (121 of 161) had a baseline pathogen isolated from the infection site. All pathogens isolated at baseline were susceptible to JNJ-Q2. All Gram-positive pathogens were susceptible to linezolid; linezolid susceptibility testing against Gram-negative organisms was not checked. Investigators were allowed to withdraw patients from the study if they grew Gram-negative organisms and were not improving, but none chose to do so. A total of 104 patients grew S. aureus pathogens; 63% of these patients grew MRSA (38 MRSA isolates from 36 patients in the JNJ-Q2 group and 32 MRSA isolates from 29 patients in the linezolid group). Seventy of 139 baseline pathogens were identified as MRSA.
Early treatment response data were analyzed using the approaches noted in . In the primary efficacy analysis based on the SAP criteria for early clinical response, 62/83 (74.7%) and 62/78 (79.5%) patients in the JNJ-Q2 and linezolid groups, respectively, were considered to have achieved success; JNJ-Q2 did not meet criterion for noninferiority under the SAP (odds ratio [OR], 0.758; 95% confidence interval [CI], 0.361-1.594), i.e., the lower bound of the 95% CI needed to be greater than 0.469 to conclude noninferiority. In post hoc analyses based on the FDA end point criteria, JNJ-Q2 was statistically noninferior to linezolid (). A total of 61.4% of patients in the JNJ-Q2 group were considered treatment responders, compared with 57.7% of patients in the linezolid group (RD, 3.8%; 95% CI, −12.7 to 20.2%; P = 0.024).
Early responder results (lack of spread of lesion and afebrile) for the ITT population
In prespecified, secondary analyses, JNJ-Q2 had a higher percentage of clinical response, as determined by the investigator, at day 8, TOC, and SFU and was also statistically noninferior to linezolid for clinical response at all 3 time points () for the ITT population. At SFU, 83.1% of patients (69 of 83) in the JNJ-Q2 group and 82.1% (64 of 78) of ITT patients in the linezolid group were assessed as cured. Additionally, for the diabetic subgroup 12/14 (two patients were identified as cured but were evaluated outside the SFU window of 2 to 14 days posttreatment) who received JNJ-Q2 and 7/7 in the linezolid group were identified as cured. Similar findings were noted for the mITT, the CE, and the ME populations. (Note that the CE and ME populations were identical.)
Median infection site area decreased from baseline to SFU for both treatments (). Although lesion size measurements were highly variable in both treatment groups at baseline, the variation was notably larger in the linezolid group at the day 8, TOC, and SFU visits. Improvement was observed for all signs and symptoms of infection in both treatment groups, as indicated by a decrease in symptom scores, starting at day 2. Marked improvement was observed for the signs and symptoms recorded from day 2 to day 4 and from day 4 to day 8.
Fig. 1. Improvement in lesion size at primary infection by visit. Box-whisker legend: box, 25th percentile, median, and 75th percentile. Whisker ends are the 5th and 95th percentiles. Values greater than Whisker ends are displayed individually. Some patients (more ...)
The additional post hoc analysis that combined the early lesion assessment with a later assessment of clinical cure demonstrated similar results between the treatment groups based on the 36- to 84-h primary analysis lesion responder window (JNJ-Q2, 59.0%; linezolid, 57.7%) as well as based on the more narrow 48- to 72-h FDA lesion responder window (JNJ-Q2, 50.6%; linezolid, 41.0%).
The clinical cure rate for patients with MRSA was higher for JNJ-Q2 (50.0% [18 of 36]) compared with the linezolid group (44.8% [13 of 29]) at day 8 (). The clinical cure rate for patients with MRSA was more than 60% at TOC (OR, 0.942 (0.335–2.652) and at SFU were 80.6% and 86.2% for JNJ-Q2 and linezolid, respectively (OR, 0.664 [0.173–2.546]).
Clinical response by visit and by baseline pathogen
At the SFU visit, approximately 84% of patients in both treatment groups had a microbiological response of presumed or confirmed eradication. Four patients (two per treatment group) had a microbiological response of presumed persistence. No patient had a microbiologically confirmed persistent pathogen.
Microbiologically confirmed posttreatment new infection (isolation of a new pathogen from a posttreatment culture, with signs and symptoms of infection) was observed in one subject in each group (JNJ-Q2, Staphylococcus epidermidis; linezolid, Enterobacter cloacae and Escherichia coli). Long-term clinical follow-up data collected 12 weeks after the last dose of medication showed similarly low rates of microbiologically confirmed recurrent infections for both treatment groups: 1.2% (n = 1) for JNJ-Q2 (MRSA isolate) and 2.6% (n = 2) for linezolid (MRSA in one subject and methicillin-susceptible S. aureus in one subject). However, it must be emphasized that the definitions used for “new infections” and “recurrent infections” were overly conservative, in that the definitions did not require the same location as the index lesion, nor did the definition require the use of additional antibiotics. Importantly, only one linezolid patient required additional antibiotics.
Clinical responses at the TOC visit for patients with a polymicrobial infection are noted in , while clinical responses at TOC for patients that grew a baseline pathogen by infection type (wound infection, abscess, and cellulitis) are presented in . Similar outcomes for clinical cure were noted between the two arms for both of these approaches.
Clinical response rates for polymicrobial infections at TOC visit
Clinical response rates at TOC visit by type of infection in patients who grew at least one baseline pathogen
There were several key clinical signs of infection considered in the current study, including fever, C-reactive protein (CRP), white blood cell (WBC) count, and lymphadenopathy. Body temperature measurements were similar between treatment groups, with a mean baseline body temperature of 36.7°C for both groups, which showed little mean change over time. At baseline, more than 80% of patients (131 of 161) had CRP levels higher than the ULN (5 mg/liter) and >45% of patients (77/161) had a CRP >5 times the ULN. At TOC more than 75% of patients in both groups had CRP levels within the normal range. Of the 68 and 63 patients that had elevated CRP values at baseline in the JNJ-Q2 and linezolid groups, respectively, 44 (65%) and 49 (78%) reported normal levels at SFU. At baseline, 22/83 (26.5%) and 26/78 (33.3%), respectively, for JNJ-Q2 and linezolid had elevated (>10,000/mm3) white blood cell levels, and by SFU there were 5 patients on JNJ-Q2 and 9 patients on linezolid with elevated white cell counts. Seventy (84%) of JNJ-Q2 and 64 (82%) of linezolid patients reported some lymphadenopathy at baseline; by SFU, 80 (96%) and 77 (99%) of JNJ-Q2 and linezolid patients, respectively, reported no lymphadenopathy. Clinical responses at the TOC visit based on presence of baseline local and systemic findings are presented in .
Clinical response rates at TOC relative to baseline local and systemic findings
The incidence, relationship to study medication, seriousness, and severity of AEs and the incidence of AEs of special interest were similar between the JNJ-Q2 and linezolid groups, except nausea and vomiting, which were greater in the JNJ-Q2 group. In the JNJ-Q2 group, 50 patients (60.2%) experienced a total of 111 AEs, of which three events in 2 patients (3.6%) were considered severe. In the linezolid group, 51 patients (64.6%) experienced a total of 110 AEs, of which five events in 4 patients (5.1%) were considered severe.
The most common AEs (>5% of patients overall) were nausea, diarrhea, vomiting, and headache. The incidences of nausea and vomiting were higher in the JNJ-Q2 group than the linezolid group. For patients who experienced nausea and vomiting in the JNJ-Q2 group, ≥80% of patients experiencing these events had resolution of their symptoms within the first 2 days of therapy; no patients discontinued treatment due to these symptoms, which were typically mild. Adverse events occurring in at least 5% of patients are listed in .
Adverse events of special interest for quinolones were prospectively monitored. Patients with diarrhea were evaluated for Clostridium difficile and no cases were identified. All rashes (<4% and with similar rates across both treatments) were considered grade 1/2 based on the DAIDS AE grading table. One subject (JNJ-Q2 group) discontinued due to a nonserious AE of allergic dermatitis. Rates of headache and dizziness were comparably recorded across both treatments. There were no other signals for CNS or other neurological adverse events in patients receiving JNJ-Q2; specifically, there were no AEs of visual disturbances or seizure disorders.
Laboratory and other investigations of special interest for quinolones were also closely monitored. There were no clinically significant AEs or SAEs of hematologic abnormalities, hyperglycemia, or hypoglycemia in patients receiving JNJ-Q2. Overall, liver chemistries were similar between the two treatment groups: 7/83 patients (8.4%) and 7/79 patients (8.9%) in the JNJ-Q2 and linezolid treatment groups, respectively, demonstrated the combination of at least a 1.5-fold rise above the ULN for ALT and at least a 1.5-fold increase above baseline in ALT. However, one JNJ-Q2 subject who had an elevated baseline ALT (61 U/liter, with ULN of 33U/liter) was found, on routine monitoring at day 8 (at the end of her treatment course with JNJ-Q2), to have an asymptomatic ALT elevation to 875 without concomitant increase in bilirubin. This finding was reported as an AE and resolved to the normal range by 30 days postrandomization. Additionally, mild, transient, asymptomatic lipase elevations were observed in 2 patients (1 in each treatment group).
AEs of asymptomatic, slight prolongation of the QTc interval, without cardiovascular sequelae were reported for 2 patients receiving JNJ-Q2 and 2 patients receiving linezolid. At the SFU visit (time of detection) both JNJ-Q2 patients had been off drug for at least 2 days: one had a baseline QTc of 450 ms that increased to 492 ms, while the other had a baseline value of 437 ms that increased to 459 ms. Similarly, one subject in the linezolid group had a baseline value of 430 ms that increased to 470 ms (day 8), while the other had a baseline of 444 ms that increased to 450 ms on day 13.
In the JNJ-Q2 group, 2 patients experienced SAEs. One subject experienced a road traffic accident with multiple injuries and compartment syndrome approximately 8 weeks after completing therapy. The second subject was reported to have an SAE of sepsis. However, the diagnosis was based on suspicion of bacteremia; only 1 of 4 predose blood cultures demonstrated a Gram-positive organism on Gram stain, but all four cultures failed to grow any bacteria. The sepsis was interpreted by the investigator to be present prior to therapy. In the linezolid group, 3 patients experienced 4 SAEs: cellulitis with allergic dermatitis requiring hospitalization, cholecystitis, and supraventricular tachycardia.