P values should not merely be used to categorize results into significant and non-significant. This practice disregards clinical relevance, confounds non-significance with no effect and underestimates the likelihood of false-positive results. Better than to use the P value as a dichotomizing instrument, the P values and the confidence intervals around effect estimates can be used to put research findings in a context, thereby taking clinical relevance but also uncertainty genuinely into account.
Olaf M Dekkers
Herman Verloop, Johannes W A Smit, and Olaf M Dekkers
Thyroid function abnormalities are common during treatment with tyrosine kinase inhibitors such as sorafenib. Suggested causes are direct effects on thyroid tissue and increased extrathyroidal metabolism of serum thyroxine and 3,5,3-triiodothyronine. We postulated that tyrosine kinase inhibitors may affect the peripheral metabolism of TSH as well. The effect of sorafenib on TSH clearance was studied.
In a study of athyreotic patients on TSH suppression therapy, TSH concentrations were measured after recombinant human TSH (rhTSH) injections before and after 26 weeks of sorafenib therapy.
Before and after the last week of sorafenib therapy, 20 patients with progressive differentiated thyroid carcinoma received a standard dose regimen of two injections 0.9 mg rhTSH on two consecutive days. TSH concentrations were measured 48 h (TSH48 h) and 96 h (TSH96 h) after the first rhTSH injection. The area under the curve (TSH-AUC), reflecting TSH content between 48 and 96 h following rhTSH administration, was calculated.
TSH48 h levels (120.5 mU/l before vs 146.3 mU/l after; P=0.029), TSH96 h levels (22.0 mU/l before vs 35.5 mU/l after; P=0.001), and TSH-AUC (142.7 vs 186.8 mU/l; P=0.001) were significantly higher after sorafenib treatment. Higher sorafenib doses were associated with increased changes in TSH96 h and TSH-AUC. In two patients, TSH levels after sorafenib therapy exceeded 200 mU/l.
Sorafenib therapy is accompanied by higher rhTSH levels, probably due to a decreased TSH clearance. Further studies are recommended to clarify whether a decreased clearance of TSH is sorafenib specific.
Natasha M Appelman-Dijkstra, Marnick Rijndorp, Nienke R Biermasz, Olaf M Dekkers, and Alberto M Pereira
Recombinant human growth hormone (rhGH) replacement is advocated in adult growth hormone-deficient (GHD) patients to increase bone mass and improve lipid profile, body composition, and quality of life. The long-term effects of discontinuation of rhGh replacement are unknown.
This cohort study and systematic review aim to evaluate the long-term metabolic effects of discontinuation of rhGh replacement in adult GHD patients, with a subgroup analyses according to age (< or > 60 years). Data on anthropometry, lipids, glucose, and bone mass density (BMD) were assessed for 3 years after discontinuation.
Cohort study included 64 patients who had discontinued rhGh replacement for >12 months. Fat percentage increased from 31.5±9.5% to 33.8±9.0% (mean difference 2.3, P=0.003). BMI decreased only in subjects <60 years (P=0.014). Glucose, total cholesterol, and LDL-cholesterol levels did not change; however, the percentage of patients on statins increased slightly from 39% to 44%. HDL-C concentration increased only in patients <60 years (mean difference 0.2, P=0.043). Lumbar spine BMD did not change; however, femoral neck BMD and bone turnover markers decreased in subjects <60 years (P=0.001). Systematic review included eight studies (n=166 patients) with a follow-up duration of 6–18 months. Of the eight studies, three qualified as low risk of bias and five as having an intermediate risk of bias. None of the studies reported handling of statins, bisphosphonates, and glucose-lowering medication or excluded patients using these medications.
In this study, discontinuation of rhGh replacement resulted in metabolic changes only in patients <60 years after 3 years. Further research warrants to determine the optimal strategies for (dis)continuation of rhGh replacement in adult patients with GHD.
Charlotte Steffensen, Alberto M Pereira, Olaf M Dekkers, and Jens Otto L Jørgensen
Type 2 diabetes (T2D) and Cushing’s syndrome (CS) share clinical characteristics, and several small studies have recorded a high prevalence of hypercortisolism in T2D, which could have therapeutic implications. We aimed to assess the prevalence of endogenous hypercortisolism in T2D patients.
Systematic review and meta-analysis of the literature.
A search was performed in SCOPUS, MEDLINE, and EMBASE for original articles assessing the prevalence of endogenous hypercortisolism and CS in T2D. Data were pooled in a random-effect logistic regression model and reported with 95% confidence intervals (95% CI).
Fourteen articles were included, with a total of 2827 T2D patients. The pooled prevalence of hypercortisolism and CS was 3.4% (95% CI: 1.5–5.9) and 1.4% (95 CI: 0.4–2.9) respectively. The prevalence did not differ between studies of unselected patients and patients selected based on the presence of metabolic features such as obesity or poor glycemic control (P = 0.41 from meta-regression). Imaging in patients with hypercortisolism (n = 102) revealed adrenal tumors and pituitary tumors in 52 and 14% respectively.
Endogenous hypercortisolism is a relatively frequent finding in T2D, which may have therapeutic implications.
Esther Donga, Olaf M Dekkers, Eleonora P M Corssmit, and Johannes A Romijn
The aim of this study was to perform a systematic review and meta-analysis on insulin resistance in adult patients with type 1 diabetes mellitus compared to healthy controls, assessed by hyperinsulinemic euglycemic clamp studies.
Design and methods
We conducted a systematic search of publications using PubMed, EMBASE, Web of Science and COCHRANE Library. Hyperinsulinemic euglycemic clamp studies comparing adult patients with type 1 diabetes mellitus to healthy controls were eligible. Primary outcome measures were pooled mean differences of insulin sensitivity of endogenous glucose production (EGP), of glucose uptake and of lipolysis. We estimated mean (standardized) differences and 95% CIs using random effects meta-analysis.
We included 38 publications in this meta-analysis. The weighed mean differences in EGP during hyperinsulinemia between patients and controls was 0.88 (95% CI: 0.47, 1.29) in the basal state and 0.52 (95% CI: 0.09, 0.95) in insulin stimulated conditions, indicating decreased hepatic insulin sensitivity in patients. Insulin sensitivity of glucose uptake was either reported as M value (M), glucose infusion rate (GIR), glucose disposal rate (GDR) or metabolic clearance rate (MCR). Weighed mean differences were similar for M −3.98 (95% CI: −4.68, −3.29) and GIR −4.61 (95% CI: −5.86, −3.53). Weighed mean difference for GDR was −2.43 (95% CI: −3.03, −1.83) and −3.29 (95% CI: −5.37, −1.22) for MCR, indicating decreased peripheral insulin sensitivity in patients. Insulin mediated inhibition of lipolysis was decreased in patients, reflected by increased non-esterified fatty acid levels.
Insulin resistance is a prominent feature of patients with type 1 diabetes mellitus and involves hepatic, peripheral and adipose tissues.
Rolf H H Groenwold and Olaf M Dekkers
The results of observational studies of causal effects are potentially biased due to confounding. Various methods have been proposed to control for confounding in observational studies. Eight basic aspects of confounding adjustment are described, with a focus on correction for confounding through covariate adjustment using regression analysis. These aspects should be considered when planning an observational study of causal effects or when assessing the validity of the results of such a study.
Wiebke Arlt, Olaf M Dekkers, Juliane Léger, and Robert K Semple
Olaf M Dekkers, Pia Burman, and On behalf of the ESE Clinical Committee
Herman Verloop, Olaf M Dekkers, Robin P Peeters, Jan W Schoones, and Johannes W A Smit
Iodothyronine deiodinases represent a family of selenoproteins involved in peripheral and local homeostasis of thyroid hormone action. Deiodinases are expressed in multiple organs and thyroid hormone affects numerous biological systems, thus genetic variation in deiodinases may affect multiple clinical endpoints. Interest in clinical effects of genetic variation in deiodinases has clearly increased. We aimed to provide an overview for the role of deiodinase polymorphisms in human physiology and morbidity. In this systematic review, studies evaluating the relationship between deiodinase polymorphisms and clinical parameters in humans were eligible. No restrictions on publication date were imposed. The following databases were searched up to August 2013: Pubmed, EMBASE (OVID-version), Web of Science, COCHRANE Library, CINAHL (EbscoHOST-version), Academic Search Premier (EbscoHOST-version), and ScienceDirect. Deiodinase physiology at molecular and tissue level is described, and finally the role of these polymorphisms in pathophysiological conditions is reviewed. Deiodinase type 1 (D1) polymorphisms particularly show moderate-to-strong relationships with thyroid hormone parameters, IGF1 production, and risk for depression. D2 variants correlate with thyroid hormone levels, insulin resistance, bipolar mood disorder, psychological well-being, mental retardation, hypertension, and risk for osteoarthritis. D3 polymorphisms showed no relationship with inter-individual variation in serum thyroid hormone parameters. One D3 polymorphism was associated with risk for osteoarthritis. Genetic deiodinase profiles only explain a small proportion of inter-individual variations in serum thyroid hormone levels. Evidence suggests a role of genetic deiodinase variants in certain pathophysiological conditions. The value for determination of deiodinase polymorphism in clinical practice needs further investigation.
Olaf M Dekkers and Rolf H H Groenwold
Immortal time bias should always be considered in an observational study if exposure status is determined based on a measurement or event that occurs after baseline. This bias can lead to an overestimation of an effect, but also to an underestimation, which is explained. Several approaches are illustrated that can be used to avoid immortal time bias in the analysis phase of the study; a time-dependent analysis to avoid immortal time bias optimizes the use of available information.