Agnès Dechartres

Efficacy and safety of rituximab in common variable immunodeficiency-associated immune cytopenias: a retrospective multicentre study on 33 patients.

Patients with common variable immunodeficiency (CVID) are at high risk of developing immune thrombocytopenia (ITP) and/or autoimmune haemolytic anaemia (AHA). Given their underlying immunodeficiency, immunosuppressive treatment of these manifestations may increase the risk of infection. To assess efficacy and safety of rituximab in patients with CVID‐associated ITP/AHA, a multicentre retrospective study was performed. Thirty‐three patients, 29 adults and four children, were included. Patients received an average of 2·6 treatments prior to rituximab including steroids, intravenous immunoglobulin and splenectomy (21%). The median ITP/AHA duration at time of first rituximab administration was 12 months [range 1–324] and the indication for using rituximab was ITP (22 cases), AHA (n = 5) or both (n = 7); 1 patient was treated sequentially for ITP and then AHA. The overall initial response rate to rituximab was 85% including 74% complete responses. After a mean follow‐up of 39 ± 30 months after rituximab first administration, 10 of the initial responders relapsed and re‐treatment with rituximab was successful in 7/9. Severe infections occurred after rituximab in eight adults (24%), four of whom were not on immunoglobulin replacement therapy. In conclusion, rituximab appears to be highly effective and relatively safe for the management of CVID‐associated severe immune cytopenias.

Preresection serum C-reactive protein measurement and survival among patients with resectable non-small cell lung cancer

OBJECTIVE This study aimed to determine whether preresection serum CRP level independently predicts survival among patients with resectable non-small cell lung cancer. METHODS Clinical, pathologic, and laboratory data from 300 patients operated on for non-small cell lung cancer in a single institution were studied in univariate and multivariate survival analyses. Validation was sought in another cohort of 68 similar patients from another institution. RESULTS In the main cohort, preoperative CRP value was 3 mg/L or lower in 136 patients (45.3%), between 4 and 20 mg/L in 89 (29.7%), and greater than 20 in 64 (21.3%). CRP level was significantly associated with chronic bronchitis, hypoalbuminemia, pathologic stage, and peritumoral vascular emboli. Overall, 5-year survivals of patients with preoperative CRP 3 mg/L or lower, between 4 and 20 mg/L, and greater than 20 mg/L were 55.6%, 45.6%, and 40.0%, respectively (P = .0571). In multivariate analysis, CRP level greater than 20 was significantly associated with survival, but with significant interaction between CRP level and disease stage (P = .02). Patients in stage I or II disease with CRP levels greater than 20 had worse survival than did patients with undetectable CRP (adjusted hazard ratio, 1.874; 95% confidence interval, 1.039-3.381); the difference was not significant in stages III and IV. In the validation series, CRP level greater than 20 mg/L also predicted worse survival (P = .018). CONCLUSIONS Preoperative CRP level greater than 20 mg/L is significantly associated with worse survival than undetectable CRP in patients with stage I or II non-small cell lung cancer.

Outcome and prognostic factors of pleural mesothelioma after surgical diagnosis and/or pleurodesis

OBJECTIVE The objective of this study was to evaluate long-term survival and prognostic factors in patients with malignant pleural mesothelioma. METHODS All consecutive patients referred for surgical diagnosis and/or pleurodesis for malignant pleural mesothelioma between 2000 and 2010 were studied. The following parameters were prospectively recorded: age, sex, tobacco consumption, asbestos exposure, type and duration of symptoms, American Society of Anesthesiologists (ASA) score, body mass index, preoperative C-reactive protein levels, white blood cells and platelet count, pachypleuritis on chest radiograph, type of diagnostic surgical procedure, histologic type, modality of pleurodesis, and chemotherapy. Survival was assessed on March 1, 2011. RESULTS A total of 170 patients were included. For the entire population, median survival was 12 months (95% confidence interval [CI], 10-15). Two-, 5-, and 7-year overall survival was 26% (95% CI, 19-35), 11% (95% CI, 6-21), and 5% (95% CI, 9-22), respectively. Asbestos exposure, age, ASA class III versus ASA classes I and II, nonepithelioid histology, C-reactive protein levels >3 mg/L, and white cell count >12,000/mm(3) influenced outcome in univariate analysis. Multivariate analysis showed that nonepithelioid histology (hazard ratio [HR], 2.76; 95% CI, 1.50-5.08); age (HR, 1.05; 95% CI, 1.01-1.08); C-reactive protein levels between 4 and 50 mg/L, and >51 (HR, 2.28; 95% CI, 1.18-4.42; and HR, 2.69; CI, 1.29-5.60, respectively); and leukocytosis >12,000/mm(3) (HR, 2.28; 95% CI, 1.22-4.25) were independent worse survival predictors. CONCLUSIONS Median survival in an unselected population of patients with malignant pleural mesothelioma treated nonsurgically is 12 months. Nonepithelioid histology, older age, abnormal C-reactive protein levels, and leukocytosis are independent predictors of worse survival.

Sharing of Data From Industry-Funded Registered Clinical Trials

Sharing of Data From Industry-Funded Registered Clinical Trials Access to individual patient-level data from clinical trials could be an important step forward in clinical research.1-4 Some pharmaceutical companies have committed to share such data. The largest repository is the Clinical Study Data Request (CSDR) website, wherein companies voluntarily list studies for which data can be requested.5 To evaluate the completeness of data sharing on CSDR, we investigated the proportion of randomized clinical trials (RCTs) registered at ClinicalTrials.gov that were listed at CSDR.

Effect of Atropine With Propofol vs Atropine With Atracurium and Sufentanil on Oxygen Desaturation in Neonates Requiring Nonemergency Intubation: A Randomized Clinical Trial

Importance Propofol or a combination of a synthetic opioid and muscle relaxant are both recommended for premedication before neonatal intubation but have yet to be compared. Objective To compare prolonged desaturation during neonatal nasotracheal intubation after premedication with atropine-propofol vs atropine-atracurium-sufentanil treatment. Design, Setting, and Participants Multicenter, double-blind, randomized clinical trial (2012-2016) in 6 NICUs in France that included 173 neonates requiring nonemergency intubation. The study was interrupted due to expired study kits and lack of funding. Interventions Eighty-nine participants were randomly assigned to the atropine-propofol group and 82 to the atropine-atracurium-sufentanil group before nasotracheal intubation. Main Outcomes and Measures The primary outcome was prolonged desaturation (SpO2 <80% lasting > 60 seconds), using intention-to-treat analysis using mixed models. Secondary outcomes assessed the characteristics of the procedure and its tolerance. Results Of 173 neonates randomized (mean gestational age, 30.6 weeks; mean birth weight, 1502 g; 71 girls), 171 (99%) completed the trial. Of 89 infants, 53 (59.6%) in the atropine-propofol group vs 54 of 82 (65.9%) in the atropine-atracurium-sufentanil group achieved the primary outcome (adjusted RD, −6.4; 95% CI, −21.0 to 8.1; P = .38). The atropine-propofol group had a longer mean procedure duration than did the atropine-atracurium-sufentanil group (adjusted RD, 1.7 minutes; 95% CI, 0.1-3.3 minutes; P = .04); a less frequent excellent quality of sedation rate, 51.7% (45 of 87) vs 92.6% (75 of 81; P < .001); a shorter median time to respiratory recovery, 14 minutes (IQR, 8-34 minutes) vs 33 minutes (IQR, 15-56 minutes; P = .002), and shorter median time to limb movement recovery, 18 minutes (IQR, 10-43 minutes) vs 36 minutes (IQR, 19-65 minutes; P = .003). In the 60 minutes after inclusion, SpO2 was preserved significantly better in the atropine-propofol group (time × treatment interaction P  = .02). Of the atropine-propofol group 20.6% had head ultrasound scans that showed worsening intracranial hemorrhaging (any or increased intraventricular hemorrhage) in the 7 days after randomization vs 17.6% in the atropine-atracurium-sufentanil group (adjusted RD, 1.2; 95% CI, −13.1 to 15.5, P = .87). Severe adverse events occurred in 11% of the atropine-propofol group and in 20% of the atropine-atracurium-sufentanil group. Conclusions and Relevance Among neonates undergoing nonemergency nasotracheal intubation, the frequency of prolonged desaturation did not differ significantly between atropine used with propofol or atropine used with atracurium and sufentanil. However, the study may have been underpowered to detect a clinically important difference, and further research may be warranted. Trial Registration ClinicalTrials.gov Identifier: NCT01490580, EudraCT number: 2009-014885-25

Early acute respiratory distress syndrome after pneumonectomy: Presentation, management, and short- and long-term outcomes

Objective Postpneumonectomy acute respiratory failure leading to invasive mechanical ventilation carries a severe prognosis, especially when acute respiratory distress syndrome develops. The aim of this study was to describe the risk factors, management, and outcome of postpneumonectomy acute respiratory distress syndrome. Methods We retrospectively reviewed the clinical files of patients undergoing pneumonectomy in a single center between 2005 and 2015. Risk factors for acute respiratory distress syndrome, management characteristics, and short‐ and long‐term outcomes were assessed. Results Among the 543 patients undergoing pneumonectomy, 89 (16.4%) needed reintubation within the 30th postoperative day, including 60 (11%) who developed acute respiratory distress syndrome. At multivariate analysis, right‐side pneumonectomy (odds ratio [OR], 2.75; 95% confidence interval [CI], 1.51‐5.02; P = .0009) and higher Charlson Comorbidity Index (OR, 1.26; 95% CI, 1.07‐1.49; P = .007) were identified as independent risk factors for acute respiratory distress syndrome. Operative mortality was 8.1% for all pneumonectomies, 43.8% (n = 39/89) in intubated patients, and 56.7% (34/60) in patients with acute respiratory distress syndrome. Mortality was higher in severe (25/36, 69.4%) than in mild or moderate acute respiratory distress syndrome (9/24, 37.5%, P = .014). Logistic regression identified 3 independent predictors of operative mortality in patients with acute respiratory distress syndrome: age (OR, 1.08; 95% CI, 1.01‐1.15; P = .02), right pneumonectomy (OR, 5.97; 95% CI, 1.33‐26.71; P = .02), and severe acute respiratory distress syndrome (OR, 7.19; 95% CI, 1.74‐29.73; P = .006). Five‐year survival was 17.6% for patients with acute respiratory distress syndrome. Conclusions Acute respiratory distress syndrome is a severe early complication of pneumonectomy with a poor outcome. The low survival underlines the need for novel management strategies.

Patients experiencing early acute respiratory failure have high postoperative mortality after pneumonectomy.

Objective: Post‐pneumonectomy acute respiratory failure leading to invasive mechanical ventilation carries a severe prognosis especially when acute respiratory distress syndrome occurs. The aim of this study was to describe risk factors and outcome of acute respiratory failure. Methods: We retrospectively reviewed clinical files of all patients who underwent pneumonectomy in a single center between 2005 and 2015. Risk factors and outcome of acute respiratory failure were assessed in univariate and multivariate analysis. Results: Among the 543 patients who underwent pneumonectomy in the period of study, 89 (16.4%) needed reintubation within the 30th postoperative day and 60 of these (11% of all pneumonectomies) developed acute respiratory distress syndrome. In multivariate analysis, right‐side of pneumonectomy (odds ratio [OR], 2.29; 95% confidence interval [CI], 1.24–4.22), chronic cardiac disease (OR, 2.15; 95% CI, 1.08–4.25), Charlson Comorbidity Index (OR, 1.35; 95% CI, 1.14–1.61), carinal resection (OR, 3.23; 95% CI, 1.26–8.29), and extrapleural pneumonectomy (OR, 8.36; 95% CI, 3.31–21.11) were identified as independent risk factors of reintubation. Thirty‐day mortality was 7.7% for all pneumonectomies, 41.6% (37/89) in the invasive ventilation group, and 53.3% (32/60) in patients with acute respiratory distress syndrome. In non‐reintubated patients, 30‐day mortality was 1.1% (5/454). In reintubated patients, 5‐year survival was 27.1% (95% CI, 17.8–41.4). Conclusions: Early acute respiratory failure requiring reintubation remains a severe complication of pneumonectomy with a poor outcome.

Association Between Publication Characteristics and Treatment Effect Estimates: A Meta-epidemiologic Study

Systematic reviews of randomized controlled trials (RCTs) aim to synthesize existing evidence on the efficacy and safety of health interventions (1, 2). Dissemination of findings may be influenced by many factors, including the direction and strength of results (35). Previous studies showed that trials with statistically significant results are more likely to be publishedand published more quicklythan those with negative results (3, 6, 7), leading to dissemination bias. To prevent this bias from affecting the results of systematic reviews and meta-analyses, methodological standards recommend extensive literature searches, including searches for unpublished studies in trial registries, regulatory documents, and conference proceedings (810). Retrieving unpublished data uses time and resources (1113); whether the effort is worth the value remains controversial (1416). Such data are frequently incomplete (17), and risk of bias is difficult if not impossible to evaluate (15). In addition, conference proceedings may report preliminary or intermediate results, which may be misleading (18, 19). It is generally believed that searches should not be restricted to the English language (8), but evidence on the association between publication language and treatment effect estimates is conflicting. In 1998, Egger and colleagues (20) showed that German authors were more likely to publish RCTs in English if results were statistically significant, which suggests an English-language bias. In contrast, 2 meta-epidemiologic studies suggested larger treatment effects in trials published in non-English languages than in English (4, 21, 22). These studies were published many years ago, before widespread use of the Internet, which modified publication patterns. Most Cochrane reviews search for nonEnglish-language publications and unpublished data (11, 23, 24), although these represent a low proportion of included studies (15, 23). Using a large sample of Cochrane reviews, we aimed to compare treatment effects between published and unpublished RCTs and between trials published in languages other than English and English. Methods We did a meta-epidemiologic study using a large sample of Cochrane reviews. This study was based on a protocol developed on May 2017 (available on request) and was not registered in PROSPERO. Data Sources We initially obtained data from the 2796 intervention reviews published in the Cochrane Database of Systematic Reviews between March 2011 and September 2014. Data were provided as 1 XML file per review and included all elements entered by review authors in RevMan, the software developed by Cochrane for preparing and maintaining its reviews. A review file contained information on each included study, including methods summary, references, risk-of-bias assessment, and results (that is, number of events in each group and number of participants analyzed for binary outcomes). Using the XML package in R, version 3.2.2, we combined all individual XML files in a single database. For the purpose of this study, we manually updated this sample by searching the Cochrane Database of Systematic Reviews via PubMed for reviews published up to 1 January 2017. Study Selection Selection of Systematic Reviews We excluded withdrawn or empty (that is, including no study) reviews and those including fewer than 3 studies. We focused on reviews of RCTs only and excluded those with observational or nonrandomized studies. We excluded overviews, umbrella reviews, network meta-analyses, and reviews focusing only on continuous outcomes. We then selected reviews reporting a risk-of-bias assessment for the following items, which we aimed to take into account in sensitivity analyses: sequence generation, allocation concealment, blinding, and incomplete outcome data. For the update, we also excluded systematic reviews for which we already included a meta-analysis in the initial sample. Selection of Meta-analyses and Trials Within Reviews From eligible Cochrane reviews, we identified all comparisons that included at least 1 meta-analysis of binary outcomes with 3 or more RCTs because a meta-epidemiologic analysis requires at least 3 trials. We manually selected comparisons between an experimental intervention and placebo, usual care, or no intervention because of uncertainty about the direction of bias for comparisons of 2 active interventions. We also excluded comparisons that corresponded to sensitivity analyses. From each eligible comparison, we selected the first meta-analysis of binary outcomes that included at least 3 RCTs. We did not consider meta-analyses of adverse events because of uncertainty about the direction of bias. We considered subgroup analyses only when the forest plot reported a combined estimate of treatment effects across subgroups. When meta-analyses overlapped across comparisons (within or across reviews), defined as sharing 3 or more trials, we selected the meta-analysis with the most trials. If they included the same number of trials, we selected the first meta-analysis reported if the overlap was across comparisons within the same review and the one included in the most recent review if the overlap was across reviews. We included all trials that were included in the selected meta-analyses. We relied on data extracted by review authors, and for each trial we extracted the medical condition, interventions in experimental and control groups, outcome evaluated in the meta-analysis, year of publication, sample size, references as reported in the review, and results (number of events and number of participants analyzed in each group, as reported in the forest plot for the selected meta-analysis). We also extracted the risk-of-bias assessment for each of the following items from the Cochrane Risk of Bias Tool: sequence generation, allocation concealment, blinding, and incomplete outcome data. According to the Cochrane Handbook for Systematic Reviews of Interventions (8), blinding and incomplete outcome data should be assessed at the outcome level. Therefore, for reviews reporting an assessment of these items by outcome or type of outcomes, we manually identified the outcome evaluated in the selected meta-analysis and chose the corresponding risk-of-bias assessment. Evaluation of Publication Characteristics For each trial, we identified the primary reference as reported by the review authors. If several or no primary references were reported, we manually checked references to identify which one reported the main results of the trial. We combined approaches to evaluate the type of primary reference reported for each trial. First, we scrutinized the journal and number of pages for each primary reference to identify any trials not reported as full journal articles. We also evaluated whether the review authors relied on published or unpublished data, as reported in the references of included studies. Finally, we screened the characteristics of included studies reported by review authors, including support of judgment for the risk-of-bias assessment, for any information about the type of data they used (such as abstract only, thesis, or company document). We searched the reference in Google Scholar whenever necessary and, when in doubt, checked all other references reported for that trial. We classified each primary reference as a journal article, conference abstract, thesis, personal communication, company document, report (such as World Health Organization report), registry information, book, or unpublished data without precision. Articles in press were considered journal articles, but we systematically checked that they had actually been published. A single reviewer (C.R.) did this evaluation, and a senior reviewer (A.D.) checked the initial sample and did the evaluation for the update. The reviewers discussed all doubts to reach a consensus. Definition of Unpublished Data Unpublished trials were considered those for which results were available only as conference abstracts, personal communications, books, theses, reports, company documents, or information from clinical trial registries (such as ClinicalTrials.gov). This definition is consistent with previous studies (4, 12). Evaluation of Characteristics of Journal Articles For trials with a journal article as the primary reference, we evaluated whether this reference was indexed in MEDLINE by searching MEDLINE via PubMed with the title, authors, journal, and year of publication. We also evaluated language of publication by searching the National Library of Medicine catalog or the journal Web site for articles not indexed in MEDLINE. Data Synthesis and Analysis Descriptive analysis involved frequencies and percentages for qualitative variables and medians and interquartile ranges (IQRs) for quantitative variables. We compared treatment effect estimates between published and unpublished trials and between published trials and those available only as a conference abstract. For trials published as a journal article, we compared treatment effect estimates by MEDLINE status (not indexed vs. indexed) and publication language (language other than English vs. English). The same analytic strategy was used for all comparisons. We used odds ratios (ORs) as the measure of treatment effect, which is common in meta-epidemiologic analyses of binary outcomes because of the symmetrical properties of ORs. Outcome events were recoded so that an OR less than 1 indicated a beneficial association with the experimental intervention. The difference in treatment effect estimates was expressed as a ratio of ORs (ROR). For the comparison between published and unpublished trials, the ROR was the OR in published trials to that in unpublished trials, so an ROR less than 1 indicates larger treatment effect estimates in published than unpublished trials. We applied the 2-step method, also called the metameta-analytic approach, described by Sterne and colleagues (25). In the first step, we estimated the ROR

The association between Fibroblast Growth Factor 23 and renal transplantation outcome is modified by follow-up duration and glomerular filtration rate assessment method

Introduction Fibroblast growth factor 23 (FGF23) could contribute to cardiovascular morbidity in chronic kidney disease. In studies of kidney transplant recipients, a high circulating level of FGF23 has been associated with death and graft loss independently of estimated glomerular filtration rate (GFR). Whether FGF23 is associated with adverse outcomes in the early posttransplantation period is unknown. Methods We analyzed a cohort of 845 kidney transplant recipients in stable condition who had GFR measured in the first years after transplantation with a median follow-up of 71 months. Results A high FGF23 concentration was associated with death or graft loss in univariate analysis, but this association was lost after adjustment for measured GFR. In contrast, FGF23 remained significantly associated with the composite outcome when estimated GFR was substituted for measured GFR. We also observed that follow-up duration modified the association between FGF23 and outcome. Although FGF23 was not associated with any endpoint in the full duration of the study, we found an independent association between FGF23 and the incidence of graft loss within the 4 years after FGF23 measurement. We did not find an association between FGF23 levels and left ventricular mass in a subgroup of 227 patients who had echocardiography performed within 3 months of FGF23 measurement. Discussion This study demonstrates that FGF23 measured during the first year after transplantation is not an independent predictor of death and graft loss and is not associated with left ventricular hypertrophy in the posttransplantation period. It further unveils important factors modifying the association between FGF23 and outcome in this population.