Insulin resistance in polycystic ovary syndrome: a systematic review and meta-analysis of euglycaemic–hyperinsulinaemic clamp studies

In the paper

“This approach identified three clear outliers (Micic et al., 2007;Yang et al., 2013;Wu et al., 2014).”

• Also stated in: § Publication bias and outliers · plain

  “The other two studies were excluded from the meta-analysis (Supplementary Fig. S1).”

What the sources say

• ?Not checked — no full text of the sourcecites Micic et al., 2007 source
• Supported

  The citation identifies Yang et al. (2013) as one of the studies the citing authors classified as outliers. Our copy’s Results section reports Yang et al.’s own ZAG study; the outlier classification, correction, and exclusions are the citing authors’ meta-analytic decisions, not findings attributed to Yang et al. (LLM, conf=0.86)

  cites Yang et al., 2013 source
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Wu et al., 2014

In the paper

“IR (defined as reduced insulin sensitivity [IS] compared with control women) and hyperandrogenism play key roles in the aetiology (Teede et al., 2007(Teede et al., , 2010)).”

What the sources say

• ?Not checked — no full text of the sourcecites Teede et al., 2007 source
• Supported

  The Introduction section, “Aetiology: insulin resistance and hyperandrogenism,” explicitly identifies hyperandrogenism as an established contributor to PCOS aetiology and insulin resistance as a pathophysiological contributor. It notes that insulin resistance occurs in around 50%–80% of women with PCOS, so the claim does not imply that it is present in every case. (LLM, conf=0.94)

  cites Teede et al., 2010 source

In the paper

“Similarities in IR may be related to genetic variations that are preserved across these criteria (Chen et al., 2011;Day et al., 2015;Hayes et al., 2015), suggesting a homogenous disease origin linked to insulin-related genes (Shi et al., 2012).”

What the sources say

• ?Not checked — no full text of the sourcecites Chen et al., 2011 source
• Partly supported

  The Methods (Follow-up studies; page unavailable) compare PCOS cases defined by Rotterdam and NIH criteria, and the Results (Genome-wide association signals for PCOS; page unavailable) report directionally concordant genetic associations with no significant heterogeneity by case definition. The Results (Mendelian randomization analyses; page unavailable) also link genetically predicted insulin resistance to PCOS risk, but do not show that similarities in insulin resistance across the two diagnostic groups arise from the same variants. The sentence cites Shi et al., not Day et al., for the separate suggestion of a homogeneous disease origin. (LLM, conf=0.86)

  cites Day et al., 2015 sourceExperimental — accuracy not yet measured
• Partly supported

  The Discussion (page unavailable) reports a PCOS susceptibility locus found in both the NIH-phenotype cohort and a Chinese cohort that included additional Rotterdam phenotypes. It does not link that shared locus to similarities in insulin resistance (IR); it says the NIH phenotype has the highest IR risk and emphasizes gonadotropin-related genes. The sentence cites Shi et al., not Hayes et al., for the separate suggestion about insulin-related genes. (LLM, conf=0.84)

  cites Hayes et al., 2015 sourceExperimental — accuracy not yet measured

In the paper

“Compared with their respective controls, lean and overweight women with PCOS showed lower IS with large and very large magnitudes respectively (Table IV, Fig. 3).”

In the paper

“Low levels of sex hormonebinding globulin (SHBG) were associated with reduced levels of IS (-10%, ±10%; small, very likely negative), which was not confounded by BMI.”

• Also stated in: § IS, SHBG and hormone concentrations · plain

  “The mean lower concentration of SHBG in PCOS compared with controls was associated with a small lower IS, which was not confounded by BMI.”

• Also stated in: § Discussion · boosted

  “SHBG had a strong negative association with IR, which was not confounded by BMI.”

In the paper

“Compared with their respective controls, younger and older women with PCOS showed lower or much lower IS, respectively (large and very large magnitudes).”

In the paper

“FSH concentrations did not differ between PCOS and controls.”

In the paper

“Increasing age adversely impacted IR, yet the impact was trivial when adjusted for BMI and diagnostic criteria.”

In the paper

“Testosterone means were higher in PCOS and were associated with worse IS but the relationship indicates women with less testosterone are more insulin resistant at the same BMI.”

In the paper

“Older women diagnosed with PCOS had a lower IS compared with younger women with PCOS (a large magnitude), but this effect became small and unclear when BMI and diagnostic criteria were included in the model (Table IV).”

In the paper

“while FSH had a trivial relationship with only minor differences in FSH concentration between PCOS and control women.”

In the paper

“The unexplained residual differences between studies, represented by the random effect, were large (~21%) but fell to moderate, when all predictors were included in the model (Hopkins et al., 2009).”

What the sources say

• ?Not checked — no full text of the sourcecites Hopkins et al., 2009 source

In the paper

“We also confirm that SHBG is not only lower, but also less variable in PCOS, compared with controls (Jayagopal et al., 2003), with larger studies now needed to derive specific SHBG cut-off ranges to predict IR.”

What the sources say

• ?Not checked — no full text of the sourcecites Jayagopal et al., 2003 source

In the paper

“We report that PCOS has an intrinsic IR with a 27% lower IS, independent of BMI, with results that are statistically qualitatively 'large' and 'most likely', as defined here.”

• Also stated in: § Discussion · plain

  “Overall mean -27, ±6 Large****”

In the paper

“Here, we address these gaps and advance the field by confirming in a robust systematic review and meta-analysis, that PCOS is underpinned by an intrinsic IR (~27% reduction in IS), independent of BMI and age, as well as across different diagnostic criteria and ethnic groups.”

In the paper

“Consensus is yet to be achieved in lean women with PCOS as IR is not consistently demonstrated (Diamanti-Kandarakis and Dunaif, 2012).”

What the sources say

• ?Not checked — no full text of the sourcecites Diamanti-Kandarakis and Dunaif, 2012 source

In the paper

“two independent reviewers scanned the titles, abstracts and keywords of every record retrieved.”

In the paper

“Risk of bias of the included studies was assessed by a reviewer using criteria developed a priori (Harris, 2010).”

What the sources say

• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Harris, 2010

In the paper

“Obesity is known to increase circulating androgen and insulin levels, may increase PCOS prevalence and exacerbates the clinical features of PCOS (Fig. 1; Diamanti-Kandarakis and Dunaif, 2012;Teede et al., 2013).”

What the sources say

• ?Not checked — no full text of the sourcecites Diamanti-Kandarakis and Dunaif, 2012 source
• ?Not checked — no full text of the sourcecites Teede et al., 2013 source

In the paper

“IR and hyperinsulinaemia may directly stimulate production of androgens via ovarian tissue steroidogenesis, independently of changes in gonadotrophin concentration (Diamanti-Kandarakis and Dunaif, 2012).”

What the sources say

• ?Not checked — no full text of the sourcecites Diamanti-Kandarakis and Dunaif, 2012 source

In the paper

“The systematic review adheres to the principles of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.”

• Also stated in: § Materials and Methods · plain

  “The methodology adheres to the Preferred Reporting Items for Systematic review and Meta-Analysis (PRISMA) Protocols.”

In the paper

“This suggests a more homogeneous disease and highlights the need for further research into features such as IR in women with Rotterdam diagnosed PCOS (Barber et al., 2007;Panidis et al., 2012).”

What the sources say

• ?Not checked — no full text of the sourcecites Barber et al., 2007 source
• ?Not checked — no full text of the sourcecites Panidis et al., 2012 source

In the paper

“The strength of this meta-analysis is the extensive and comprehensive literature search and the focus on studies using gold standard methods to measure IR.”

In the paper

“A meta-analysis cannot solve inherent confounding problems in included studies, which may exaggerate or underestimate effects.”

In the paper

“This is in contrast to our current understanding of the interplay between androgens and IR (Diamanti-Kandarakis and Dunaif, 2012).”

What the sources say

• ?Not checked — no full text of the sourcecites Diamanti-Kandarakis and Dunaif, 2012 source

In the paper

“The recent NIH consensus workshop on PCOS endorsed the Rotterdam criteria as the definitive diagnostic criteria (National Institutes of Health, 2012) and recommended that all studies include analyses by reproductive PCOS phenotype.”

What the sources say

• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites National Institutes of Health, 2012

In the paper

“There was a lack of available robust trials and challenges of extracting reported data from the included studies as outlined.”

In the paper

“IR can be measured using numerous direct and indirect techniques, including homeostatic model assessment, quantitative insulin sensitivity check index and oral glucose tolerance test-related measures comprising area under the curve for glucose and insulin (Muniyappa et al., 2008).”

What the sources say

• ?Not checked — no full text of the sourcecites Muniyappa et al., 2008 source

In the paper

“These are based on fasting or post glucose-load blood glucose and insulin levels, yet lack accuracy and reliability and are inadequate for clinical purposes and mechanistic research where IR is a primary study outcome (Muniyappa et al., 2008).”

What the sources say

• ?Not checked — no full text of the sourcecites Muniyappa et al., 2008 source

In the paper

“Low SHBG levels are also increased by insulin-sensitising medication (Ehrmann et al., 1997;Moghetti et al., 2000;Moran et al., 2013);”

What the sources say

• ?Not checked — no full text of the sourcecites Ehrmann et al., 1997 source
• ?Not checked — no full text of the sourcecites Moghetti et al., 2000 source
• ?Not checked — no full text of the sourcecites Moran et al., 2013 source

In the paper

“Inconsistent reporting of results made it necessary to express variables as factors by dividing the given variable in the PCOS group by that in the respective control group.”

In the paper

“The euglycaemic-hyperinsulinaemic clamp is considered the reference standard to directly measure IS in research settings and can be used to define a specific cut-off level for IR (Muniyappa et al., 2008).”

What the sources say

• ?Not checked — no full text of the sourcecites Muniyappa et al., 2008 source

In the paper

“The inability to acquire missing data from all eligible studies is not unexpected and deemed part of the meta-analysis process (Kelley et al., 2004).”

What the sources say

• ?Not checked — no full text of the sourcecites Kelley et al., 2004 source

In the paper

“Other key factors may also influence IR in PCOS, including low-grade inflammation (Shorakae et al., 2015) and sympathetic nervous system dysfunction (Diamanti-Kandarakis and Dunaif, 2012;Shorakae et al., 2015) and further research is needed to clarify key mechanisms underpinning IR in PCOS.”

What the sources say

• ?Not checked — no full text of the sourcecites Shorakae et al., 2015 source
• ?Not checked — no full text of the sourcecites Diamanti-Kandarakis and Dunaif, 2012 source

In the paper

“Small sample sizes and missing data limited evidence synthesis and meta-analysis.”

In the paper

“Also endpoints varied (including units and scales).”

In the paper

“This is consistent with mechanistic studies demonstrating that SHBG production is suppressed by IR and hyperinsulinaemia (Le et al., 2012) and that SHBG concentrations increase with insulin sensitiser treatment (Moghetti et al., 2000).”

What the sources say

• ?Not checked — no full text of the sourcecites Moghetti et al., 2000 source

In the paper

“Furthermore, not all identified studies were included because of difficulties sourcing the required data from original authors.”

In the paper

“The body of evidence was heterogeneous such that diagnostic criteria were not uniformly applied with many studies only including women with NIH-diagnosed PCOS, and few reporting reproductive phenotypes.”