Summary of key features of LTE studies
| Design | LTE duration | Analysis population | Primary outcome | Analysis | Safety reporting | Imputation of missing data in initial DB-RCT/subsequent LTE | Rescue arm in initial DB-RCT | |
|---|---|---|---|---|---|---|---|---|
| Synthetic DMARD | ||||||||
| Leflunomide 200317 | Inclusion from two phase III studies; ≥2 years leflunomide exposureOpen-label, LTE | Upon leflunomide availability (last observation) | Intent to treat | Not stated(Long-term safety and efficacy outcomes (ACR response, HAQ)) | Descriptive% Responders of all patients on continued leflunomide | AE and SAE: (frequency) | LOCF/not stated | No |
| Biological DMARD | ||||||||
| Anakinra 200212 | Blinded extensionOriginal placebo group randomly assigned to three doses in LTE23 | 1 Year | Intent to treat | ACR20, 12 months (6 months LTE) | Completers onlySeparate RCT placebo analysis | 18 Months (12 months into LTE)No (%) | LOCF/non-responder imputation; LOCF | No |
| Etanercept 200313 | Open-label extensionWeek 12: methotrexate±steroid dose reduction | Observational, on-going studyMedian 44 months drug exposure | Per protocol | Not statedMax. 3 year efficacy/safety evaluation | Completers for each annual time point (ACR) | SAE/patient-year | Non-responder imputation24/not stated | No |
| ARMADA (adalimumab) 200625 | Open-labelPI determined co-medication change after 6 months extension | Up to 4 years of adalimumab exposure (with methotrexate) | Per protocolEfficacy:2–4 year exposure4 year exposure | Not stated | 6-monthly efficacy (ACR response) | AE/100 patient-yearsObserved-expected malignancy SIR (95% CI) | LOCF5/not stated | Escape open-label option in DB-RCT study |
| Etanercept 200614 | Previous seven trials (phase I–III, open-label)Original etanercept dose year 1; licensed dose post year 1Additional DMARD/steroid post year 1 | To data cut-off time (ongoing study) | Intent to treat year 1Only completers subsequent years | SafetyPhysical functionQuality of life | Efficacy: Completers for each LTE year onlySafety: AE and SAE per 100 patient-years | Safety from phase I–III, open-label±LTENon-serious AE year 1SAE throughout LTE (no/100 patient-years)Malignancy (event/100 patient-years)Deaths/100 patient-years | None used | No |
| Rituximab 200711 | Open-labelInclusion from three previous phase IIa, IIb, III (DB-RCT/open-label) | Data from specified clinical programme | Per protocol | Safety and efficacyACR 20 week 24 after each rituximab course | Descriptive; on observed data for efficacy analysis | AE & SAEAll exposure population | Not stated | No |
| AIM 2008 (abatacept)9 | Blinding maintained in extension (original placebo arm unknown) | 2 Years | Intent to treatPrespecified analysisAs-observed data (inc post hoc) | Not statedEfficacy outcomes:ACR responseACR70 for >6 and >9 months | Responder analysisNon-completers included within non-responder analysis | Incidence rate AE, SAE, infection/100 patient-yearsFrequencies for rarer events/malignancy | Non-responder imputation/Not stated | No |
| STREAM 2009 (tocilizumab)26 | Open-label | 5 Years | Per protocol | Not statedEfficacy: parameters/exposure years | Efficacy: of completersContinuation rate | SAE/infection/malignancy/100 patient-yearsHaemoglobin/lipid: descriptive statistics | None used | No |
| Toclilzumab 3.5 year data (abstract)15 | Open-label (patients from four phase III DB-RCT)8 27–29DMARD±TNF inhibitor inadequate responder | 3.5 Years | ‘As observed’ | Efficacy outcomes:ACR 20, 50, 70 responseHAQ=0LDADAS28 remission | Baseline=first active dose of tocilizumab (ie, initial RCT baseline for active randomised group; rescue time point or LTE baseline for initial RCT placebo group)Proportion of patients achieving ACR response, HAQ=0Absolute numbers and proportion of patients LDA and DAS28 remission | Safety data reported (separately); median 3.1 years30AE/100 patient-years | None usedNo imputation of assessments (patient excluded if missing measurement)LOCF for missing TJC and SJC | Yes (all four DB-RCT) |
A summary of several recently reported long-term extension studies illustrates the variation in content, analysis and method of reporting.
ACR, American College of Rheumatology; AE, adverse event; DAS28, disease activity score; DB, double-blind; DMARD, disease modifying antirheumatic drug; HAQ, health assessment questionnaire; LDA, low disease activity; LOCF, last observation carried forward; LTE, long-term extension; PI, principal investigator; RCT, randomised controlled trial; SIR, standardised incidence ratio; SAE, serious adverse event; SJC, swollen joint count; TJC, tender joint count; TNF, tumour necrosis factor.