Nonlinear effects and effect modification at the participant-level in IPD meta-analysis part 2: methodological guidance is available

OBJECTIVES

To review methodological guidance for nonlinear covariate-outcome associations (NL), and linear effect modification and nonlinear effect modification (LEM and NLEM) at the participant level in individual participant data meta-analyses (IPDMAs) and their power requirements.

STUDY DESIGN AND SETTING

We searched Medline, Embase, Web of Science, Scopus, PsycINFO and the Cochrane Library to identify methodology publications on IPDMA of LEM, NL or NLEM (PROSPERO CRD42019126768).

RESULTS

Through screening 6,466 records we identified 54 potential articles of which 23 full texts were relevant. Nine further relevant publications were published before or after the literature search and were added. Of these 32 references, 21 articles considered LEM, 6 articles NL or NLEM and 6 articles described sample size calculations. A book described all four. Sample size may be calculated through simulation or closed form. Assessments of LEM or NLEM at the participant level need to be based on within-trial information alone. Nonlinearity (NL or NLEM) can be modeled using polynomials or splines to avoid categorization.

CONCLUSION

Detailed methodological guidance on IPDMA of effect modification at participant-level is available. However, methodology papers for sample size and nonlinearity are rarer and may not cover all scenarios. On these aspects, further guidance is needed.

Immunotherapy for people with clinically isolated syndrome or relapsing-remitting multiple sclerosis: treatment response by demographic, clinical, and biomarker subgroups (PROMISE)-a systematic review protocol

BACKGROUND

Multiple sclerosis (MS) is an inflammatory and degenerative disease of the central nervous system with an increasing worldwide prevalence. Since 1993, more than 15 disease-modifying immunotherapies (DMTs) have been licenced and have shown moderate efficacy in clinical trials. Based on the heterogeneity of the disease and the partial effectiveness of therapies, a personalised medicine approach would be valuable taking individual prognosis and suitability of a chosen therapy into account to gain the best possible treatment effect. The primary objective of this review is to assess the differential treatment effects of all approved DMTs in subgroups of adults with clinically isolated syndrome or relapsing forms of MS. We will analyse possible treatment effect modifiers (TEM) defined by baseline demographic characteristics (gender, age), and diagnostic (i.e. MRI measures) and clinical (i.e. relapses, disability level) measures of MS disease activity.

METHODS

We will include all published and accessible unpublished primary and secondary analyses of randomised controlled trials (RCTs) with a follow-up of at least 12 months investigating the efficacy of at least one approved DMT, with placebo or other approved DMTs as control intervention(s) in subgroups of trial participants. As the primary outcome, we will address disability as defined by the Expanded Disability Status Scale or multiple sclerosis functional composite scores followed by relapse frequency, quality of life measures, and side effects. MRI data will be analysed as secondary outcomes. MEDLINE, EMBASE, CINAHL, LILACS, CENTRAL and major trial registers will be searched for suitable studies. Titles and abstracts and full texts will be screened by two persons independently using Covidence. The risk of bias will be analysed based on the Cochrane "Risk of Bias 2" tool, and the certainty of evidence will be assessed using GRADE. Treatment effects will be reported as rate ratio or odds ratio. Primary analyses will follow the intention-to-treat principle. Meta-analyses will be carried out using random-effects models.

DISCUSSION

Given that individual patient data from clinical studies are often not available, the review will allow to analyse the evidence on TEM in MS immunotherapy and thus support clinical decision making in individual cases.

SYSTEMATIC REVIEW REGISTRATION

PROSPERO CRD42021279665 .

Using the Causal Inference Framework to Support Individualized Drug Treatment Decisions Based on Observational Healthcare Data

When healthcare professionals have the choice between several drug treatments for their patients, they often experience considerable decision uncertainty because many decisions simply have no single “best” choice. The challenges are manifold and include that guideline recommendations focus on randomized controlled trials whose populations do not necessarily correspond to specific patients in everyday treatment. Further reasons may be insufficient evidence on outcomes, lack of direct comparison of distinct options, and the need to individually balance benefits and risks. All these situations will occur in routine care, its outcomes will be mirrored in routine data, and could thus be used to guide decisions. We propose a concept to facilitate decision-making by exploiting this wealth of information. Our working example for illustration assumes that the response to a particular (drug) treatment can substantially differ between individual patients depending on their characteristics (heterogeneous treatment effects, HTE), and that decisions will be more precise if they are based on real-world evidence of HTE considering this information. However, such methods must account for confounding by indication and effect measure modification, eg, by adequately using machine learning methods or parametric regressions to estimate individual responses to pharmacological treatments. The better a model assesses the underlying HTE, the more accurate are predicted probabilities of treatment response. After probabilities for treatment-related benefit and harm have been calculated, decision rules can be applied and patient preferences can be considered to provide individual recommendations. Emulated trials in observational data are a straightforward technique to predict the effects of such decision rules when applied in routine care. Prediction-based decision rules from routine data have the potential to efficiently supplement clinical guidelines and support healthcare professionals in creating personalized treatment plans using decision support tools.

Pragmatic Trials in Genomic Medicine: The Integrating Pharmacogenetics In Clinical Care (I-PICC) Study

Pragmatic clinical trials (PCTs) have an established presence in clinical research and yet have only recently garnered attention within the landscape of genomic medicine. Using the PRagmatic-Explanatory Continuum Indicator Summary 2 (PRECIS-2) as a framework, this paper illustrates the application of PCT principles to The Integrating Pharmacogenetics In Clinical Care (I-PICC) Study, a trial of pharmacogenetic testing prior to statin initiation for cardiovascular disease prevention in primary care. The trial achieved high engagement with providers (85% enrolled of those approached) and enrolled a representative sample of participants for which statin therapy would be recommended. The I-PICC Study has a high level of pragmatism, which should enhance the generalizability of its findings. The PRECIS-2 may be useful in the design and evaluation of PCTs of genomic medicine interventions, contributing to the generation of evidence that can bridge the gap between genomics innovation and clinical adoption.

When drug treatments bias genetic studies: Mediation and interaction

Background

Increasingly, genetic analyses are conducted using information from subjects with established disease, who often receive concomitant treatment. We determined when treatment may bias genetic associations with a quantitative trait.

Methods

Graph theory and simulated data were used to explore the impact of drug prescriptions on (longitudinal) genetic effect estimates. Analytic derivations of longitudinal genetic effects are presented, accounting for the following scenarios: 1) treatment allocated independently of a genetic variant, 2) treatment that mediates the genetic effect, 3) treatment that modifies the genetic effect. We additionally evaluate treatment modelling strategies on bias, the root mean squared error (RMSE), coverage, and rejection rate.

Results

We show that in the absence of treatment by gene effect modification or mediation, genetic effect estimates will be unbiased. In simulated data we found that conditional models accounting for treatment, confounding, and effect modification were generally unbiased with appropriate levels of confidence interval coverage. Ignoring the longitudinal nature of treatment prescription, however (e.g. because of incomplete records in longitudinal data), biased these conditional models to a similar degree (or worse) as simply ignoring treatment.

Conclusion

The mere presence of (drug) treatment affecting a GWAS phenotype is insufficient to bias genetic associations with quantitative traits. While treatment may bias associations through effect modification and mediation, this might not occur frequently enough to warrant general concern at the presence of treated subjects in GWAS. Should treatment by gene effect modification or mediation be present however, current GWAS approaches attempting to adjust for treatment insufficiently account for the multivariable and longitudinal nature of treatment trajectories and hence genetic estimates may still be biased.

Linear regression and the normality assumption

OBJECTIVES

Researchers often perform arbitrary outcome transformations to fulfill the normality assumption of a linear regression model. This commentary explains and illustrates that in large data settings, such transformations are often unnecessary, and worse may bias model estimates.

STUDY DESIGN AND SETTING

Linear regression assumptions are illustrated using simulated data and an empirical example on the relation between time since type 2 diabetes diagnosis and glycated hemoglobin levels. Simulation results were evaluated on coverage; i.e., the number of times the 95% confidence interval included the true slope coefficient.

RESULTS

Although outcome transformations bias point estimates, violations of the normality assumption in linear regression analyses do not. The normality assumption is necessary to unbiasedly estimate standard errors, and hence confidence intervals and P-values. However, in large sample sizes (e.g., where the number of observations per variable is >10) violations of this normality assumption often do not noticeably impact results. Contrary to this, assumptions on, the parametric model, absence of extreme observations, homoscedasticity, and independency of the errors, remain influential even in large sample size settings.

CONCLUSION

Given that modern healthcare research typically includes thousands of subjects focusing on the normality assumption is often unnecessary, does not guarantee valid results, and worse may bias estimates due to the practice of outcome transformations.

Series: Pragmatic trials and real world evidence: Paper 3. Patient selection challenges and consequences

This paper addresses challenges of identifying, enrolling, and retaining participants in a trial conducted within a routine care setting. All patients who are potential candidates for the treatments in routine clinical practice should be considered eligible for a pragmatic trial. To ensure generalizability, the recruited sample should have a similar distribution of the treatment effect modifiers as the target population. In practice, this can be best achieved by including-within the selected sites-all patients without further selection. If relevant heterogeneity between subgroups is expected, increasing the relative proportion of the subgroup of patients in the heterogeneous trial could be considered (oversampling) or a separate trial in this subgroup can be planned. Selection will nevertheless occur. Low enrollment and loss to follow-up can introduce selection and can jeopardize validity as well as generalizability. Pragmatic trials are conducted in clinical practice rather than in a dedicated research setting, which could reduce recruitment rates. However, if a trial poses a minimal burden to the physician and the patient and routine clinical practice is maximally adhered to, the participation rate may be high and loss to follow-up will not be a specific problem for pragmatic trials.

PCSK9 monoclonal antibodies for the primary and secondary prevention of cardiovascular disease

BACKGROUND

Despite the availability of effective drug therapies that reduce low-density lipoprotein (LDL)-cholesterol (LDL-C), cardiovascular disease (CVD) remains an important cause of mortality and morbidity. Therefore, additional LDL-C reduction may be warranted, especially for patients who are unresponsive to, or unable to take, existing LDL-C-reducing therapies. By inhibiting the proprotein convertase subtilisin/kexin type 9 (PCSK9) enzyme, monoclonal antibodies (PCSK9 inhibitors) may further reduce LDL-C, potentially reducing CVD risk as well.

OBJECTIVES

Primary To quantify short-term (24 weeks), medium-term (one year), and long-term (five years) effects of PCSK9 inhibitors on lipid parameters and on the incidence of CVD. Secondary To quantify the safety of PCSK9 inhibitors, with specific focus on the incidence of type 2 diabetes, cognitive function, and cancer. Additionally, to determine if specific patient subgroups were more or less likely to benefit from the use of PCSK9 inhibitors.

SEARCH METHODS

We identified studies by systematically searching the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, and Web of Science. We also searched Clinicaltrials.gov and the International Clinical Trials Registry Platform and screened the reference lists of included studies. We identified the studies included in this review through electronic literature searches conducted up to May 2016, and added three large trials published in March 2017.

SELECTION CRITERIA

All parallel-group and factorial randomised controlled trials (RCTs) with a follow-up time of at least 24 weeks were eligible.

DATA COLLECTION AND ANALYSIS

Two review authors independently reviewed and extracted data. When data were available, we calculated pooled effect estimates.

MAIN RESULTS

We included 20 studies with data on 67,237 participants (median age 61 years; range 52 to 64 years). Twelve trials randomised participants to alirocumab, three trials to bococizumab, one to RG7652, and four to evolocumab. Owing to the small number of trials using agents other than alirocumab, we did not differentiate between types of PCSK9 inhibitors used. We compared PCSK9 inhibitors with placebo (thirteen RCTs), ezetimibe (two RCTs) or ezetimibe and statins (five RCTs).Compared with placebo, PCSK9 inhibitors decreased LDL-C by 53.86% (95% confidence interval (CI) 58.64 to 49.08; eight studies; 4782 participants; GRADE: moderate) at 24 weeks; compared with ezetimibe, PCSK9 inhibitors decreased LDL-C by 30.20% (95% CI 34.18 to 26.23; two studies; 823 participants; GRADE: moderate), and compared with ezetimibe and statins, PCSK9 inhibitors decreased LDL-C by 39.20% (95% CI 56.15 to 22.26; five studies; 5376 participants; GRADE: moderate).Compared with placebo, PCSK9 inhibitors decreased the risk of CVD events, with a risk difference (RD) of 0.91% (odds ratio (OR) of 0.86, 95% CI 0.80 to 0.92; eight studies; 59,294 participants; GRADE: moderate). Compared with ezetimibe and statins, PCSK9 inhibitors appeared to have a stronger protective effect on CVD risk, although with considerable uncertainty (RD 1.06%, OR 0.45, 95% CI 0.27 to 0.75; three studies; 4770 participants; GRADE: very low). No data were available for the ezetimibe only comparison. Compared with placebo, PCSK9 probably had little or no effect on mortality (RD 0.03%, OR 1.02, 95% CI 0.91 to 1.14; 12 studies; 60,684 participants; GRADE: moderate). Compared with placebo, PCSK9 inhibitors increased the risk of any adverse events (RD 1.54%, OR 1.08, 95% CI 1.04 to 1.12; 13 studies; 54,204 participants; GRADE: low). Similar effects were observed for the comparison of ezetimibe and statins: RD 3.70%, OR 1.18, 95% CI 1.05 to 1.34; four studies; 5376 participants; GRADE: low. Clinical event data were unavailable for the ezetimibe only comparison.

AUTHORS' CONCLUSIONS

Over short-term to medium-term follow-up, PCSK9 inhibitors reduced LDL-C. Studies with medium-term follow-up time (longest median follow-up recorded was 26 months) reported that PCSK9 inhibitors (compared with placebo) decreased CVD risk but may have increased the risk of any adverse events (driven by SPIRE-1 and -2 trials). Available evidence suggests that PCSK9 inhibitor use probably leads to little or no difference in mortality. Evidence on relative efficacy and safety when PCSK9 inhibitors were compared with active treatments was of low to very low quality (GRADE); follow-up times were short and events were few. Large trials with longer follow-up are needed to evaluate PCSK9 inhibitors versus active treatments as well as placebo. Owing to the predominant inclusion of high-risk patients in these studies, applicability of results to primary prevention is limited. Finally, estimated risk differences indicate that PCSK9 inhibitors only modestly change absolute risks (often to less than 1%).

Adjusting for bias in unblinded randomized controlled trials

It may not always be possible to blind participants of a randomized controlled trial for treatment allocation. As a result, estimators of the actual treatment effect may be biased. In this paper, we will extend a novel method, originally introduced in genetic research, for instrumental variable meta-analysis, adjusting for bias due to unblinding of trial participants. Using simulation studies, this novel method, "Egger Correction for non-Adherence", is introduced and compared to the performance of the "intention-to-treat," "as-treated," and conventional "instrumental variable" estimators. Scenarios considered (time-varying) non-adherence, confounding, and between-study heterogeneity. The effect of treatment on a binary endpoint was quantified by means of a risk difference. In all scenarios with unblinded treatment allocation, the Egger Correction for non-Adherence method was the least biased estimator. However, unless the variation in adherence was relatively large, precision was lacking, and power did not surpass 0.50. As a comparison, in a meta-analysis of blinded randomized controlled trials, power of the conventional IV estimator was 1.00 versus at most 0.14 for the Egger Correction for non-Adherence estimator. Due to this lack of precision and power, we suggest to use this method mainly as a sensitivity analysis.

Chemotherapy effectiveness and mortality prediction in surgically treated osteosarcoma dogs: A validation study

Canine osteosarcoma is the most common bone cancer, and an important cause of mortality and morbidity, in large purebred dogs. Previously we constructed two multivariable models to predict a dog's 5-month or 1-year mortality risk after surgical treatment for osteosarcoma. According to the 5-month model, dogs with a relatively low risk of 5-month mortality benefited most from additional chemotherapy treatment. In the present study, we externally validated these results using an independent cohort study of 794 dogs. External performance of our prediction models showed some disagreement between observed and predicted risk, mean difference: -0.11 (95% confidence interval [95% CI]-0.29; 0.08) for 5-month risk and 0.25 (95%CI 0.10; 0.40) for 1-year mortality risk. After updating the intercept, agreement improved: -0.0004 (95%CI-0.16; 0.16) and -0.002 (95%CI-0.15; 0.15). The chemotherapy by predicted mortality risk interaction (P-value=0.01) showed that the chemotherapy compared to no chemotherapy effectiveness was modified by 5-month mortality risk: dogs with a relatively lower risk of mortality benefited most from additional chemotherapy. Chemotherapy effectiveness on 1-year mortality was not significantly modified by predicted risk (P-value=0.28). In conclusion, this external validation study confirmed that our multivariable risk prediction models can predict a patient's mortality risk and that dogs with a relatively lower risk of 5-month mortality seem to benefit most from chemotherapy.