CoChat
Sign Up
Cited by 47Ovarian function and disordersAdrenal Hormones and DisordersSexual Differentiation and Disorders

Performance of mass spectrometry steroid profiling for diagnosis of polycystic ovary syndrome

Open access
Handelsman, D.J., Teede, H.J., Desai, R., Norman, R.J., Moran, L.J.·2017·Human Reproduction
DOI 10.1093/humrep/dew328
Open this paper in CoChat — free accountOpen PDF

Abstract

STUDY QUESTION: How well does multi-analyte steroid mass spectrometry (MS) profiling classify women with and without polycystic ovary syndrome (PCOS)? SUMMARY ANSWER: Our liquid chromatography MS (LC-MS) steroid profiling only minimally improves discrimination of women with and without PCOS compared with a direct testosterone immunoassay (T_IA) and the free androgen index (FAI). WHAT IS KNOWN ALREADY: Blood testosterone measured by direct (non-extraction) immunoassay overlaps between women with and without PCOS. Multi-analyte MS provides greater specificity and accuracy for steroid measurement so might improve the classification. STUDY DESIGN, SIZE, DURATION: An observational, cross-sectional study of women with PCOS (n = 152) defined by Rotterdam criteria and matched non-PCOS (n = 45) control women was conducted. PARTICIPANTS/MATERIALS, SETTING, METHODS: ) and cortisol were measured by LC-MS; T_IA and sex hormone binding globulin were measured by immunoassay; and FAI, calculated free testosterone (cFT) and total androgen index (TAI) were calculated. Classification was based on logistic regression with corresponding univariate and multivariate C-statistics. MAIN RESULTS AND THE ROLE OF CHANCE: [0.74], DHEA[0.71]) or derivatives (cFT [0.75], TAI [0.60]). For all multivariate models, the overall correct predictions (81-86%) featured high sensitivity (92-96%) but low specificity (28-43%). and substituting LC-MS steroid measurements for T_IA and FAI produced only minimal improvements in classification. LIMITATIONS REASONS FOR CAUTION: The study cohort is limited in size and only unconjugated steroids were measured. WIDER IMPLICATIONS OF THE FINDINGS: Multi-analyte steroid profiling of unconjugated circulating steroids provides only limited improvement on direct T_IA in classifying women with and without PCOS. STUDY FUNDING/COMPETING INTERESTS: None. TRIAL REGISTRATION NUMBER: N/A.

AI snapshot

Written from the paper’s full text.

TL;DR

In an observational cross-sectional study of 152 women with PCOS and 45 non-PCOS controls, multi-analyte LC-MS steroid profiling only minimally improved discrimination of PCOS compared with direct testosterone immunoassay (T_IA) and the free androgen index (FAI). Optimal multivariate models had high sensitivity but low specificity and showed no significant improvement over a single testosterone measurement by immunoassay or LC-MS.

Key findings

  • T_IA showed a significant upward bias versus LC-MS testosterone, with overall mean bias >100%, slope 1.86 (95% CI 1.59–2.18), and intercept 0.02 (−0.08 to 0.10); bias was most pronounced below 0.5 ng/ml (1.7 nmol/l), a range including >80% of samples (n = 164, slope = 1.99, 95% CI 1.85–2.13), versus >0.5 ng/ml (n = 32, slope = 1.14, 95% CI 1.0–1.29).
  • Women with PCOS had higher serum T, DHEA, A4, TAI, T_IA, cFT, FAI and E2 in the abstract, but the Results section reports higher T, DHEA, A4, T_IA, cFT, FAI and TAI and lower E2, with no differences in DHT, E1, P4, 17OHP4 or cortisol.
  • Univariate C-statistics were highest for FAI (0.89) and T_IA (0.82), compared with T (0.72), DHT (0.40), A4 (0.74), DHEA (0.71), cFT (0.75) and TAI (0.60).
  • Multivariate models gave overall correct predictions of 81–86%, with high sensitivity (92–96%) but low specificity (28–43%); substituting LC-MS steroid measurements for T_IA and FAI produced only minimal improvements, and all models showed no significant improvement over a single T_IA or LC-MS testosterone measurement.

Methods

  • Observational, cross-sectional analysis of biobanked baseline samples from three clinical trials: 152 premenopausal women with PCOS by Rotterdam criteria and 45 non-PCOS control women, aged 18–45 years and overweight or obese (BMI >25 kg/m²).
  • Serum T, DHT, DHEA, A4, E2, E1, 17OHP4, P4 and cortisol were measured by LC-MS; T_IA and SHBG were measured by immunoassay; FAI, calculated free testosterone (cFT) and total androgen index (TAI) were calculated.
  • Classification used logistic regression with univariate and multivariate C-statistics, forward stepping models minimizing log-likelihood, and Wald statistics; FAI was excluded from models containing serum T or SHBG due to collinearity.

Limitations

  • Stated by the paper: the cohort is limited in size; only unconjugated steroids were measured; samples were pooled from three different clinical studies of Australian women, not all of whom had ovarian ultrasound unless required for diagnosis; findings may be affected by participation bias and may not generalize to other PCOS populations, ethnic or geographic origins, referral patterns, or all T_IAs.
  • Evident from the text: the cross-sectional design cannot establish causality, and women with PCOS differed from controls in age, weight, BMI, waist circumference, blood pressure, FG score, hsCRP, HOMA and triglycerides, so confounding is possible.
  • Evident from the text: the abstract and Results section disagree on the direction of E2 differences between PCOS and control women.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1093/humrep/dew328

Cites

27 of 27 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 7 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    Univariate C-statistics were highest for FAI (0.89) and T_IA (0.82) compared with other androgens (T 0.72, DHT 0.40), pro-androgens (A4 0.74, DHEA 0.71) or derivatives (cFT 0.75, TAI 0.60).

    “Univariate C-statistics were highest for FAI (0.89) and T_IA (0.82) compared with other androgens (T [0.72], DHT [0.40]), pro-androgens (A 4 [0.74],DHEA[0.71]) or derivatives (cFT [0.75], TAI [0.60]).”
    • Also stated in: § Results · plain
      “For hormonal variables, the highest univariate C-statistics were FAI (0.89) and T_IA (0.82), with the remainder of other variables between 0.40 and 0.76”
    Quote in the paperSubstantial claim
  • Findingown data

    In Model 2, replacing serum T by MS with measurement by immunoassay (T_IA) marginally reduced all discrimination statistics.

    “In Model 2, replacing serum T by MS with measurement by immunoassay (T_IA) marginally reduced all of the discrimination statistics.”
    Quote in the paperNot a substantial claim
  • Findingown data

    Optimal discrimination was achieved with a combination of serum A4, DHEA and SHBG after evaluating but discarding serum T, DHT, DHEA, E2, E1, 17OHP4, P4 and cortisol.

    “That optimal discrimination was achieved with a combination of serum A 4 , DHEA and SHBG after evaluating but discarding serum T, DHT, DHEA, E 2 , E 1 , 17OHP 4 , P 4 and cortisol.”
    Quote in the paperSubstantial claim
  • Findingown data

    Serum testosterone measured by immunoassay demonstrates levels more than 100% higher than measured by LC-MS.

    “Serum testosterone by immunoassay demonstrated levels more than 100% higher than that measured by LC-MS.”
    • Also stated in: § Results · plain
      “The serum testosterone immunoassay (T_IA) had a significant upward bias (overall mean bias >100%) relative to LC-MS measurement”
    Quote in the paperSubstantial claim
  • Findingown data

    All models showed no significant improvement over a single measurement of testosterone by immunoassay or LC-MS.

    “All models showed no significant improvement over a single measurement of testosterone by immunoassay or LC-MS.”
    Quote in the paperSubstantial claim
  • May combine two claims. 11
  • Classified as probably not a substantial claim. 3

  • May combine two claims.

    Findingown data

    Model 3 replaced cFT and SHBG with FAI and marginally improved the positive predictive value (85-89%) but reduced the negative predictive value (76-47%) and specificity (43-28%) without a change in sensitivity or overall correct prediction.

    “Model 3 replaced cFT and SHBG with FAI and marginally improved the positive predictive value (85-89%) but reduced the negative predictive value (76-47%) and specificity (43-28%) without a change in sensitivity or overall correct prediction.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Model 4 combining the changes in Models 2 and 3 marginally improved the positive predictive value but reduced the negative predictive value without change in overall correct prediction.

    “Finally, combining the changes in Models 2 and 3 into Model 4, marginally improved the positive predictive value but reduced the negative predictive value without change in overall correct prediction.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Compared with controls, women with PCOS had higher serum T, DHEA, A4, T_IA, cFT, FAI and TAI and lower E2 but no differences in serum DHT, E1, P4, 17OHP4 or cortisol.

    “Women with PCOS also had higher serum T, DHEA, A 4 , T_IA, cFT, FAI and TAI and lower E 2 but no differences in serum DHT, E 1 , P 4 , 17OHP 4 or cortisol”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    For all multivariate models, overall correct predictions were 81-86%, with high sensitivity 92-96% but low specificity 28-43%.

    “For all multivariate models, the overall correct predictions (81-86%) featured high sensitivity (92-96%) but low specificity (28-43%).”
    • Also stated in: § Results · plain
      “Model 1 had good positive (85%) and negative (75%) predictive values and high sensitivity (96%) but relatively low specificity (43%) with an overall correct prediction of 84%”
    • Also stated in: § Discussion · plain
      “Overall correct 84% 81% 86% 86% Positive predictive value 85% 84% 89% 90% Negative predictive value 76% 63% 47% 56% Sensitivity 96% 92% 95% 94% Specificity 43% 43% 28% 40%”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Model 1 had good positive predictive value (85%) and negative predictive value (75%) and high sensitivity (96%) but relatively low specificity (43%) with an overall correct prediction of 84%.

    “Model 1 had good positive (85%) and negative (75%) predictive values and high sensitivity (96%) but relatively low specificity (43%) with an overall correct prediction of 84% (Supplementary data, Table SII).”
    • Also stated in: § Discussion · plain
      “Positive predictive value 85% 84% 89% 90% Negative predictive value 76% 63% 47% 56% Sensitivity 96% 92% 95% 94% Specificity 43% 43% 28% 40%”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Adding age or measures of obesity (weight, waist circumference, BMI) led to inclusion of serum E2 in the optimal classification but did not otherwise change the set of effective predictor variables.

    “Adding into the models age or measures of obesity (weight, waist circumference, BMI) led to inclusion of serum E 2 in the optimal classification but did not otherwise change the set of effective predictor variables (data not shown).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingcitation

    The present study confirms decreased or unchanged serum DHT and increased serum A4 concentrations in PCOS but without supporting that their measurement improves overall biochemical diagnosis of PCOS.

    “The present study confirms the decreased or unchanged serum DHT (Munzker et al., 2015;Stener-Victorin et al., 2010) and increased serum A 4 (Barth et al., 2010;Stener-Victorin et al., 2010;Janse et al., 2011;Keefe et al., 2014) concentrations in PCOS but without supporting that their measurement improves overall biochemical diagnosis of PCOS (O'Reilly et al., 2014).”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Munzker et al., 2015

      Not checked in the free run.

    • cites Stener-Victorin et al., 2010

      Not checked in the free run.

    • cites Barth et al., 2010

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Non-extraction testosterone assays in women are less accurate and display method-specific bias relative to liquid chromatography-mass spectrometry (LC-MS) measurements.

    “As non-extraction testosterone assays in women are less accurate and display method-specific bias relative to liquid chromatography-mass spectrometry (LC-MS) measurements (Handelsman and Wartofsky, 2013)”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Handelsman and Wartofsky, 2013

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    The findings may be affected by participation bias and may not be generalizable to all other PCOS populations, notably for different ethnic or geographic origins and different referral patterns or selection pressures.

    “Although the studies employed consistent clinical procedures, overlapping staff and the same MS-based measurements, nevertheless our findings may be affected by participation bias and may not be generalizable to all other PCOS populations, notably for different ethnic or geographic origins and different referral patterns or selection pressures.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    The findings using the direct T_IAs used in this study may not extrapolate to all other T_IAs.

    “Similarly, the findings using the direct T_IAs used in this study may not extrapolate to all other T_IAs.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductnone given

    The studies employed consistent clinical procedures, overlapping staff and the same MS-based measurements.

    “Although the studies employed consistent clinical procedures, overlapping staff and the same MS-based measurements”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductnone given

    The study cohort is limited in size and only unconjugated steroids were measured.

    “The study cohort is limited in size and only unconjugated steroids were measured.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    Contemporary clinical practice guidelines recommend conventional immunoassay-based measurements, despite their suboptimal biochemical accuracy or confirmation of PCOS diagnosis.

    “This is consistent with contemporary clinical practice guidelines (Teede et al., 2011;Legro et al., 2013;Conway et al., 2014) which recommend such conventional immunoassay-based measurements, despite their suboptimal biochemical accuracy or confirmation of PCOS diagnosis.”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Teede et al., 2011

      Not checked in the free run.

    • cites Legro et al., 2013

      Not checked in the free run.

    • cites Conway et al., 2014

      Not checked in the free run.

13 of 13

  • The paper relies on this without support.

    Direct (non-extraction) testosterone immunoassay displays overlap between women with and without PCOS, limiting its diagnostic potential.

    “However, serum testosterone measured by direct (non-extraction) immunoassays displays an overlap between women with and without PCOS limiting its diagnostic potential.”
    “Blood testosterone measured by direct (non-extraction) immunoassay overlaps between women with and without PCOS.”

  • The paper relies on this without support.

    Multi-analyte MS provides greater specificity and accuracy for steroid measurement, so might improve classification.

    “Multi-analyte MS provides greater specificity and accuracy for steroid measurement so might improve the classification.”
2 of 2

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Estradiol direction reverses between abstract and results

    Major

    The abstract says serum E2 was higher in women with PCOS, while the Results section says it was lower.

  • Not every discrimination statistic declined in Model 2

    Minor

    The Results section says replacing MS testosterone with immunoassay testosterone reduced all discrimination statistics, but Table I reports the same 43% specificity for Models 1 and 2.

  • DHEA is described as both retained and discarded

    Minor

    The discussion names DHEA as part of the optimal predictor combination, then lists it among the steroids discarded when forming that combination.

  • Model 3 metrics changed despite a claim of no change

    Minor

    Relative to Model 1, Table I shows Model 3 sensitivity changing from 96% to 95% and overall correct prediction changing from 84% to 86%. The Results section says neither changed.

4 of 4

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Handelsman, D. J., Teede, H. J., Desai, R., Norman, R. J., & Moran, L. J.
Affiliations
  • The University of Sydney (AU)
  • Monash University (AU)
  • The University of Adelaide (AU)
Corresponding author
Handelsman, D.J.
ORCID
Venue
Human Reproduction
Publisher
Oxford University Press (OUP)
Published
2016-12-19
Volume
32
Issue
2
Pages
418-422
ISSN
  • 0268-1161 (ISSN-L)
  • 1460-2350
DOI
10.1093/humrep/dew328
OpenAlex
W2565470299
Semantic Scholar
d6a57aa54be601a3d720e3e6631ad257846e70bc
Cited by
47
References
27
Citations by year
2026: 3 · 2025: 3 · 2024: 4 · 2023: 4 · 2022: 7 · 2021: 3 · 2020: 9 · 2019: 7 · 2018: 2 · 2017: 4
FWCI
5.74
Type
journal article
Language
en
Open access
Locations
  • Human Reproduction (publishedVersion)
  • PubMed (publishedVersion)
  • Adelaide Research & Scholarship (AR&S) (University of Adelaide) (submittedVersion)
  • HighWire Press Open Archive (submittedVersion)
Funders
  • South Australian Health and Medical Research Institute
  • National Health and Medical Research Council
Topics
  • Health Sciences › Medicine › Reproductive Medicine › Ovarian function and disorders
  • Health Sciences › Medicine › Endocrinology, Diabetes and Metabolism › Adrenal Hormones and Disorders
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Sexual Differentiation and Disorders
Keywords
polycystic ovary syndrome, steroid profiling, liquid chromatography-mass spectrometry, free androgen index, sex hormone-binding globulin, free testosterone, dehydroepiandrosterone, cortisol, diagnostic classification, diagnostic accuracy
MeSH terms
Adult, Androgens, Cross-Sectional Studies, Estrogens, Female, Humans, Hydrocortisone, Immunoassay, Polycystic Ovary Syndrome, Progestins, Mass Spectrometry

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-10.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-10.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review