Journal Pre-proof

Direct Measurement of 3-Glucuronide (PDG) In Dried Urine Spots by Liquid Chromatography-Mass Spectrometry to Detect OvulationHandelsman DJ, Nimmagadda R, Desai R, Handelsman TD, Whittle B, Skorupskaite K, Anderson RA, Direct Measurement of Pregnanediol 3-Glucuronide (PDG) In Dried Urine Spots by Liquid Chromatography-Mass Spectrometry to Detect Ovulation, Journal of Biochemistry and Molecular Biology, doi: 10.1016/j.jsbmb.2021.105900–>

David J Handelsman, Rama Nimmagadda, Reena Desai, Timothy D Handelsman, Belinda Whittle, Karolina Skorupskaite, Richard A Anderson

PII: S0960-0760(21)00093-5 DOI: https://doi.org/10.1016/j.jsbmb.2021.105900 Reference: SBMB 105900

To appear in: Journal of Steroid Biochemistry and Molecular Biology

Received Date: 18 February 2021 Revised Date: 17 March 2021 Accepted Date: 7 April 2021

Please cite this article as: { doi: https://doi.org/ This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

© 2020 Published by Elsevier. Direct Measurement of Pregnanediol 3-Glucuronide (PDG) In Dried Urine Spots by Liquid Chromatography-Mass Spectrometry to Detect Ovulation

Short title: Urine PDG for Ovulation Detection

1David J Handelsman, 1Rama Nimmagadda, 1Reena Desai, 1Timothy D Handelsman 2Belinda Whittle, 3Karolina Skorupskaite, 3Richard A Anderson 1Andrology laboratory, ANZAC Research Institute, University of Sydney, Concord Hospital, NSW 2139 Australia 2MyHealthTest Pty Ltd, Canberra ACT, Australia 3 MRC Centre for Reproductive Health, Queens Medical Research Institute, University of Edinburgh, Scotland, UK

*Correspondence: DJ Handelsman ANZAC Research Institute Hospital Road, Concord Hospital NSW 2139 E: djh @anzac.edu.au

Words: 3830 Abstract: 247 words References: 49 Tables: 4 Figures: 3 + 1 supplementary figure

Acknowledgements: This project received grant funding from the Australian Government (Entrepreneurs Programme - Innovation Connections: ICG000282 (proof of concept), Cooperative Research Centres Program: CRC-P57325) and the UK Medical Research Council (MR/N022556/1 to the MRC Centre for Reproductive Health) and the Wellcome Trust through Scottish Translational Medicine and Therapeutics Initiative 102419/Z/13/A. The authors would like to thank Agilent Technologies Australia for provision of an LC-MS instrument and technical support.

Disclaimer: RAA has undertaken consultancy work for Roche Diagnostics, Merck, Ferring, NeRRe Therapeutics and Sojournix Inc. BW is an employee of MyHealthTest Pty Ltd. The other authors have nothing to disclose in relation to this work. Journal Pre-proof

Abstract

Page 1 of 17

Background: Non-invasive self-testing using an objective chemical method to detect ovulation is valuable for women planning conception, practising contraception or undergoing infertility investigations or treatment.

Methods: Based on luteal phase secretion of (P4) and excretion of its major metabolite, pregnanediol glucuronide (PDG), we developed a novel direct liquid chromatography-mass spectrometry (LCMS) method to measure PDG and other steroid glucuronides in urine and in dried urine spots (DUS) without deconjugation or derivatization. Urine PDG by LCMS and immunoassay (P3G) and P4 by immunoassay with and without adjustment for creatinine were evaluated in daily first void urine samples from 10 women through a single menstrual cycle in which ovulation was confirmed by serial transvaginal ultrasound.

Results: Urine PDG with and without creatinine adjustment was stable during the follicular phase with the expected striking rise in the luteal phase peaking at 5 days after ovulation. Using a single spot urine sample (100 µL) or a DUS (<20 µL urine) and an optimal threshold to distinguish pre- from post-ovulatory samples, in ROC analysis urine PDG adjusted for creatinine accurately identified ovulation in 92% of samples was comparable with P3G immunoassay and superior to urine P4 with or without adjustment for creatinine. Extending the analysis to two or three consecutive daily samples reduced the false negative rate from 8% to 2.6% for two and 1.9% for three urine samples.

Conclusions: This method holds promise as a non-invasive self-test method for women to determine by an objective chemical method their ovulatory status.

Keywords: ovulation, pregnanediol glucuronide, fertility, human, female, urine

Introduction

Human fertility requires release (ovulation) and fertilisation of an oocyte by sperm delivered through patent fallopian tube(s). These basic requirements (ovulation, sperm delivery, patent tube(s)) are routinely evaluated in couples undergoing fertility investigations. Evaluating ovulatory status is initially based on a history of regular menses potentially supplemented by characteristic mid-cycle vaginal mucus changes and biphasic basal body temperature monitoring, which are reliable methods to verify ovulation (1) but require reliable daily self-testing. However, biochemical confirmation is also sought, and is essential when menstrual cycles are irregular or when pharmacological ovulation induction is used. BiochemicalJournal confirmation of ovulation is based on midPre-proof-luteal phase serum progesterone (2) or urinary excretion of conjugated metabolites of ovarian (pregnanediol 3-glucuronide (PDG) and/or 3-glucuronide (EG)) measured by immunoassay (3-5). These urinary tests supersede traditional chemical methods pioneered in the 1950’s (6, 7), which were mainly deployed in natural fertility awareness methods of contraception (8). However, verifying ovulation has wider application for women

Page 2 of 17 planning to become pregnant without needing formal medical assessment so that simpler, non-invasive tests suitable for home use are desirable.

Progesterone (P4) is a universal steroidogenic precursor (9) produced in all steroidogenic tissues which release low but detectable levels into the circulation. It is secreted in large amounts by the corpus luteum following ovulation and from the placenta during pregnancy. In non-pregnant humans, elevated serum P4 concentrations are specific for the post-ovulatory luteal phase of the menstrual cycle leading to its wide use to confirm recent ovulation. However, this requires a venepuncture by a health care professional in the mid-luteal phase of the menstrual cycle so is not readily adaptable for self-testing at home.

Alternative methods to measure the major urine metabolite of P4, pregnanediol (PD), have been developed based on measurement of PDG, the phase II (glucuronidated) excretory metabolite of PD in urine. The temporal profiles of serum P4 and urine PDG both peak at 7-8 days after ovulation so they are interchangeable biomarkers of ovulation(4). After the first chemical methods to measure urine PDG developed in the 1950’s (6), numerous immunoassay methods have been reported (10, 11). Originally these required cumbersome 24 hour urine collections, which was simplified by adapting to spot urine collections (12) including an adaptation to measurement from dried pH strips (13).

Urine PDG immunoassays have the generic methodological limitations of direct (non-extraction) immunoassays, notably cross-reactivity from structurally related steroids and urine matrix interference (14). Gas chromatography-mass spectrometry methods to measure non-conjugated PD have been available since the 1960s (15, 16) and are used in anti-doping laboratories as an endogenous reference compound (17) but current methods require pre-assay enzymatic deconjugation, solvent extraction and derivatization. Similarly, an early liquid chromatography-mass spectrometry (LC-MS) method to measure urinary excretion of sex steroid metabolite conjugates (18) did not include PDG whereas the first mass spectrometry (MS) method to measure urine PDG used an older, obsolete hard ionization technique (19). More recent LC-MS methods are reported to measure PDG in methanol-extracted serum (20) and in unextracted urine, a matrix that impairs column performance and lifetime (21), but not in dried urine spots.

We therefore developed a method to measure PDG by LC-MS in dried urine spots (DUS) without deconjugation or derivatization suitable for adaptation to a home self-test modality and avoids need for cold storage and transporting liquid urine samples. DUS sampling technology has been used in screening for drugs and intermediary metabolites (22-26), and most recently for unconjugated (27) and (28), but there are no reports of application to steroid glucuronides. As controls, we aimed to measure excretion of additional urinary steroid glucuronides, (T) 17-glucuronide (G), (DHT) 17-G, 5α,3α-androstanediol (5α 3-G, (A) 3-G and (Etio) 3-G to investigate their possible menstrual cycle variation. Journal Pre-proof

Page 3 of 17

Materials and Methods

Materials

Steroid glucuronides (T 17-G (catalog # D507c), DHT 17-G (S006), 5α 3-G (S004), A 3-G (D572) and Etio 3- G (D607) and internal standards (IS) (d3-T-G (D507a), d3-DHT-G (D573b), d4-A-G (D829f), d5-Etio-G (S020)) were all purchased from National Measurement Institute (NMI), Sydney, Australia. Other 13 chemicals of analytical grade used include PDG (P6040-000, Steraloids, Newport RI) and C5-PDG (CLM- 10412, Cambridge Isotopes, Andover, MA), methanol (Merck), ammonium fluoride and formic acid (Aldrich) and synthetic urine (Surine, Novachem).

Participants

Ethical approval was given by the South East Scotland Research Ethics Committee (Ref: 09/S1101/67) and all women gave written informed consent. First void daily urine samples were provided by 10 women (n=262 samples) through a single full menstrual cycle and were stored at -20 C until measurement as described previously (29, 30). The day of ovulation was defined by the appearance and disappearance of a dominant follicle in serial transvaginal ultrasound performed every 2-3 days (31). In the original study, urine PDG was measured by an in-house ELISA immunoassay as described (29) and is termed P3G here to distinguish the LCMS measurement of the same analyte in the present study. In addition, urine P4 was also measured by a direct progesterone immunoassay (nmol/L, Abbott Architect c8000, Abbott Laboratories) and creatinine (µmol/L) by a specific autoanalyzer enzymatic method (29) with the data reproduced here for comparison.

Steroid glucuronide measurement

Urine samples, standards, and quality control samples (100 µl) were pipetted into 96 well (1 ml) plates followed by addition of 300 µl cold methanol fortified with stable isotope labelled internal standards 13 (d3-T-G, d3-DHT-G, d4-A-G, d5-Etio-G, C5 PDG).. The mixture was sonicated for 10 minutes followed by centrifugation at 2000 rpm for 5 minutes and freezing the samples at -20C for 1.5 hours. Subsequently, the organic solvent extract (50 µl) was transferred onto another plate well containing 200 µl of 2 mM ammonium fluoride followed by 40 µl injection into the UPLC column. For lower steroid concentrations (<10 nmol/L), a larger sample (200 µl) was extracted with the extract reconstituted in 100 µl of 10% methanol:90% 2 mM ammonium fluoride followed by mixing on an orbital shaker and injection of 40 µl into the UPLC. Processed samples were then run at 35 C on an Agilent 1290 infinity II UPLC system for analysis of urine and Shimadzu Nexera 30A for DUS analysis comprising an Agilent Poroshell 120EC-C18 column (50 x 2.1 x 2.7 mm, #699775-902) equipped with a C18 guard column (Phenomenex UHPLC C18 2.1 mm, #AJO-B8782). The binary UPLC pump gradient comprised buffer A (2 mM ammonium fluoride in milliQ water) and buffer B (methanol) in 7 segments (90%A (0.5 min, 0.5 ml/min) -> 55% A (1.0 min, 0.5 ml/min) -> 40%A (6 min, 0.5 ml/min) -> 0%A (7 min, 0.5 ml/min) -> 0%A (9 min, 0.8 ml/min) -> 90%A (9.5 min, 0.8 ml/min) -> 90%A (10 min, 0.5 ml/min) followed by wash cycle comprising sequential 95:5 acetonitrileJournal (Merck): isopropyl alcohol (Honeywell) , Pre-proof0.1% formic acid and distilled water. The total run time was 10 minutes per sample. The separation achieved is shown in Supplementary Figure 1.

Urine samples emerging from UPLC were directed into an Agilent 6495B mass spectrometer equipped with an Agilent Jet Stream Electrospray ionization source (gas 230 C, gas flow 14 L/min) applying multiple reaction monitoring (MRM) scan in negative ionization mode. Capillary voltage was -4000V, nozzle voltage negative 1500 V, sheath gas flow of 7 L/min, sheath gas heater at 250 C, nebuliser gas

Page 4 of 17 pressure 15 psi, ifunnel RF voltage (high pressure -150 V, low pressure -60 V), electron multiplier voltage negative (0 V) and cycle time 2 sec for each time segment. The MRM transitions monitored as qualifier 13 and quantifier (marked as *) ions with their collision energies were PDG (495-> 85,75*, 35), C5-PDG (500-> 85,75*, 35), T-17-G (463 -> 85,75*, 35), d3-T-G (466-> 85,75*, 35), DHT 17-G (465-> 85,75*, 50), A-3-G (465-> 85,75*,35), d4-A-G (469 -> 85,75*, 35), Etio-3-G (465-> 85,75*, 50), d5-Etio-G (470- >85,75*, 35). The sensitivity (limit of detection (LOD) and limit of quantitation (LOQ) defined according to the FDA (32) and EMEA(33)), within-run and between-run (3-day) QCs reproducibility at 3 levels (accuracy and CV%) and extraction recovery (6 replicates at 3 levels on pooled urine samples) were PDG (2 nmol/L, 10 nmol/L; 92-108%, 92-113%, 5-16%, 6-17%; 95-106%), T-G (4 nmol/L, 40 nmol/L;99-109%, 98-109%, 2-12%, 6-10%; 91-106%), DHT-G (4 nmol/L, 40 nmol/L; 85-107%, 97-106%, 2-18%, 6-16%; 80- 104%), 5α-G (2 nmol/L, 10 nmol/L; 85-111%, 98-103%, 5-21%, 6-16%; 97-114%), A-G (4 nmol/L, 40 nmol/L; 87-103%, 86-106%, 2-13%, 6-10%; 92-105%), Etio-G (4 nmol/L, 40 nmol/L; 85-108%, 85-106%, 5- 12%, 5-8%; 92-111%). Conversion factors for SI units are: PDG 1 ng/ml = 2.01 nmol/L; T-G 1 ng/ml = 2.15 nmol/L; DHT-G 1 ng/ml = 2.14 nmol/L; 5α-G 1 ng/ml = 2.13 nmol/L; A-G ng/ml = 2.14 nmol/L ; Etio-G 1 ng/ml = 2.14 nmol/L. Due to the unavailability of a stable isotope of 5α-G, d3-T-G was used as the IS for 5α-G. For each steroid glucuronide, the linear reportable range was 10-10,000 nmol/L using an 8-point quadratic fit (r>0.99) standard curve. Samples outside the calibration range and were repeated with 50 uL of sample for urine analysis.

For DUS analysis the mass spectrometry conditions comprised Sciex 6500 triple quadrupole mass spectrometer (Applied Biosystems/MDS SCIEX, Foster City, CA/Concord, Ontario, Canada) equipped with an electrospray ionization (ESI) source and operated in negative ion mode. Nitrogen was employed as curtain, nebulizer, collision, and lamp gases. Multiple reaction monitoring (MRM) was used to quantify the steroids. Settings for the various transitions were optimized by infusing pure standard material into the mass spectrometer. Unit mass resolution was set in both mass-resolving quadrupole Q1 and Q3. Nitrogen and zero grade air were supplied using a PEAK A320DR/NM20ZL unit (Peak Scientific Instruments, Renfrewshire, Scotland). Nitrogen was used for curtain gas (20), ion source gas 1 (55), ion source gas 2 (50) and collision gas (6). The ESI probe temperature was 450 °C, declustering potential (- 80), entrance potential and CXP (-10) and the ion spray voltage was set to -4500 V. Steroids were quantified by multiple reaction monitoring (MRM) using settings for the various transitions optimized by infusing pure steroid into the mass spectrometer. Unit mass resolution was used in both mass-resolving quadruples Q1 and Q3. One qualifier and quantifier ion were optimized for each analyte.

The MRM transitions monitored as qualifier and quantifier (marked as *) ions with their collision 13 energies were PDG (495-> 75*, -45), C5-PDG (500-> 75*, -45), A-3-G (465-> 85,75*, -40), d4-A-G (469 -> 85,75*, -40, Etio-3-G (465-> 85,75*, -40), d5-Etio-G (470->85,75*, -40). The sensitivity (limit of detection (LOD) and limit of quantitation (LOQ) defined according to the FDA (32) and EMEA(33)), within-run and between-run (3-day) QCs reproducibility at 3 levels (50, 500 and 2000 ng/ml; accuracy, CV%) and extraction recovery on pooled urine samples (6 replicates at 3 levels, 50, 500 and 2000 ng/ml) were PDG (10 nmol/L,Journal 20 nmol/L; 94-102%, 97-103%, 5-8%, 7-10Pre-proof%; 93-104%),, A-G (10 nmol/L, 20 nmol/L; 3- 5%,86-95%, 87-102%, 5-7%; 89-113%), Etio-G (10 nmol/L, 20 nmol/L; 4-7%, 91-98%, 88-105%, 5-7%; 90- 107%). For each steroid glucuronide, the linear reportable range was 10-10,000 nmol/L using an 8-point quadratic fit (r>0.99) standard curve. Samples outside the calibration range and were further diluted 1:2 for DUS analysis.

Page 5 of 17

For urine and DUS analysis the carryover in blank injections after the highest calibrant was insignificant for all the analytes. Comparison of spiked samples with neat solution for each analyte did not show any notable matrix effects. Interference was suspected if the peak area ratio of quantifier to qualifier transitions deviated more than ±20% from the mean ion ratio calculated from the standards but none was detected.

Dried urine spots (DUS)

Urine or synthetic urine (Surine, Novachem) alone or spiked with calibration standards and QC standards were spotted onto filter paper (Ahlstrom 226 filter paper 3’x2.5’-9, Perkin Elmer), dried for 24 hr at room temperature and then stored at 4 C with desiccant until use. For extraction of DUS, 200 µl of standards, QC samples and samples were spotted onto each card and air dried. Then three 6 mm circular punches (impregnated with a mean 5.8 µl urine) were placed in a 1 ml deep 96 well plate (Storplat-96v, Perkin Elmer) to which 200 µl 2 mM ammonium fluoride was added. The plate was tightly sealed with Parafilm and shaken for 30 min on a plate shaker after which 300 µl of methanol (including internal standard) was added, again shaken for 30 mins on a plate shaker. Next the plate was centrifuged (2500 rpm, 10 min) with samples in the plate then frozen (-80 C) for 30mins. Then 75 µl of the solvent was transferred to 75 µL of 2 mM ammonium fluoride with 50 µl injected into the LC-MS. Recovery of DUS PDG measurements over the full range of standards (10-4,000 nmol/L) was stable for 2 months with filter papers stored at room temperature (90–121%) and for 7 days at 50 C (84-113%).

Data analysis

Data were reported as mean and standard error of mean for continuous data and median with inter- quartile range (Q1, Q3) otherwise. Comparison of urine and DUS measurements of the same urine samples was performed by Passing-Bablok regression and Bland-Altman deviation analyses in NCSS2020 and MedCal software. Outlier detection was by Grubbs test for extreme studentized deviates (ESD).

For analysis, the daily urine steroid concentrations were re-centred on the day of ovulation as defined by serial transvaginal ultrasound. The optimal threshold for distinguishing between pre-ovulatory and post-ovulatory specimens was identified by linear discriminant analysis with each day of the cycle classified as ovulatory (day of ovulation and later) or pre-ovulatory (days prior to ovulation) and confirmed by the peak Youden index in the receiver operating characteristic (ROC) analysis.

The window of detection was estimated as the number of days from the first day to the last day that a woman’s urine PDG with or without adjustment for creatinine exceeded the threshold.

Test diagnostic accuracy was evaluated by plotting ROC curves with the true positive (sensitivity) rate on y axis and false positive rate (1-specificity) on the x axis. Youden’s index was calculated as the sum of Y=sensitivity + specificity -1. The overall test accuracy was determined by the area-under-curve (AUC) and distance of Youden’s index to the corner with higher and lower values, respectively, indicating superior testJournal performance as well as the likelihood andPre-proof diagnostic odds ratios. Alternative testing strategies for applying this urine PDG adjusted for creatine were evaluated for 1, 2 or 3 consecutive samples exceeding the optimal threshold. This was done by direct enumeration of the data as well as by a bootstrap resampling simulation comprising 5000 random samples redrawn with

Page 6 of 17 replacement from the serial daily urine PDG/creatinine measurements. For 2 or 3 samples, the results were considered positive or negative if all samples were positive or negative.

Results

Urine PDG, P3G and P4 were detectable in all urine samples. PDG measurement in urine and DUS was highly correlated with DUS displaying no significant difference in slope, intercept, or mean bias (table 1, figure 1). Measurements of Etio-G and A-G in the same urine and DUS samples also showed high correlation but with minor variations in slope for A-G, intercept for Etio-G and mean bias due higher measurements in DUS. One extreme outlier in PDG analyses (ESD 11.9) was deleted from method comparisons. The PDG immunoassay (P3G) display significant deviation from PDG by LCMS representing concentration-dependent over-estimation (figure 2). Replicating a previously described method of using urine-impregnated pH strips (13, 34) displayed suboptimal reproducibility for PDG measurement, partly due to matrix effects (data not shown).

The daily urine PDG in urine or DUS (figure 2) and P4 (supplementary figure 2, redrawn for comparison from (29)) concentrations without and with adjustment for creatinine are shown over a full menstrual cycle centred on the day of ovulation. There was no significant change in urine PDG or P4 without or with adjustment for creatine during the follicular phase (prior to day of ovulation) except for urine PDG adjusted for creatinine which significantly increased (p=0.02) on the day before ovulation (figure 2). As control variables, the daily profiles of urine 5α-G, A-G and Etio-G showed no significant (p>0.7) increase related to ovulation (data not shown). Neither T-G nor DHT-G were consistently detectable in the urine samples.

The performance of PDG, P3G and P4 with and without adjustment for creatinine in identifying recent ovulation are shown in the ROC analysis (figure 3) and analytically in table 2. According to the AUC in the ROC analysis (table 3), PDG adjusted for creatinine was not significantly different from the PDG immunoassay but both were superior to P4 immunoassay. The optimal thresholds and performance of the different measurements of diagnostic accuracy for urine PDG, P3G and P4 with and without adjustment for creatinine are shown in table 2.

Using an optimized threshold of urine PDG adjusted for creatinine, recent ovulation was confirmed in a single urine specimen with 92% accuracy with positive predictive value 93% and negative predictive value 92%. Alternative testing strategies of more than a single urine sample evaluated by comparison with two or three samples on consecutive days are shown in table 3. For a single urine sample, the average window of detection was 10 days per cycle (table 4). Journal Pre-proof

Page 7 of 17

Discussion

The present study describes a novel direct LC-MS method to measure PDG in dried urine spot specimens without deconjugation or derivatization using methanol-diluted urine. When adjusted for urine creatinine, the utility of measuring urinary PDG to verify ovulation was evaluated by applying it to daily urine samples collected by ten women through a full ovulatory menstrual cycle during, which the day of ovulation was identified by serial transvaginal ultrasound examinations. Using an optimized threshold of urine PDG adjusted for creatinine and only a single urine specimen, recent ovulation was confirmed with 92% accuracy and creating an average window of detection of 10 days per cycle. The 8% false negative and 8% false positive results from a single urine specimen were reduced when the same method was applied to consecutive daily urine samples. Using two consecutive daily samples false negative and false positive rates were reduced to 2.6 and 5.3%, respectively and for three consecutive daily urine specimens these were further reduced to 1.9% and 3.6%. When adjusted for urinary creatinine, the performance of this method in ROC analysis was comparable to that of a PDG immunoassay (29)

The present findings using a novel LC-MS method applied to urine specimens to detect ovulation is consistent with previous findings measuring urine PDG using immunoassays. Direct comparison of the present LCMS measurements with the same analyte by an in-house ELISA immunoassay demonstrates concentration-dependent upward bias of the immunoassay, consistent with many previous studies of immunoassay performance compared with reference LCMS methods (35-38). Nevertheless, the overall performance of this LCMS method was comparable with that of the same analyte by an immunoassay but extend it to application in dried urine spots. The comparable performance is likely to represent cross-reactivity of the immunoassay with structurally related steroids (including precursors and metabolites of progesterone) that increase with progesterone during the luteal phase of the menstrual cycle. Urine samples used in this study had been stored frozen for prolonged periods but sex steroids are stable in frozen solution for 10 years (39) and specifically urine PDG is stable during frozen storage for at least 24 weeks (40), making it unlikely that the duration of frozen storage influenced the findings.

The present study relied upon an independent determination of the day of ovulation by serial ultrasound (31), now standard for research studies of hormonal determinants and outcomes of ovulation. Ideally such studies would perform daily or second daily transvaginal ultrasound within the tolerable limits of this invasive procedure. In this study scans were performed every 2-3 days so some misjudgement of the actual day of ovulation could not be excluded. Such misclassification may explain the small rise in urine PDG adjusted for creatinine on the designated day before ovulation.

Urine is an excretory product without any fixed composition so that standardization of analytical measurements in urine is unavoidable to avoid “noise” from different times of collection and hydration states nullifying interpretation of urine measurements. Furthermore, 24-hour urine collections are cumbersome and impractical for routine clinical application so most recent studies focus on spot urine collections (12). Hence as spot urine samples comprise urine accumulated over various time intervals and hydration/dilutionJournal states, previous studies of urinePre-proof PDG have suggested adjustments based on time of urine collection(41), urine specific gravity(42) or creatinine(43) in order to improve interpretation of urine measurements. However, accurate estimation of collection time of a spot urine sample depends on subjective and potentially faulty recall and does not account for hydration state. Urine specific gravity is an objective measure accounting for hydration state but is not useful if urine is very dilute(42). Urine creatinine adjustment is also objective, based on the body’s stable daily creatinine output and

Page 8 of 17 inherently adjusts for time of collection and hydration state. Although urine creatinine adjustment has limitations when applied to epidemiological studies over a wide range of women’s ages (because the body’s daily creatinine production declines with age (43)), that is not relevant to the limited age range of reproductively active women nor to use with serial samples from an individual woman. The present study confirms that for considering a single individual’s serial urine samples, adjustment for creatinine reduces misclassification error and improves diagnostic accuracy of ovulation.

In this study a single urine sample was highly effective in identifying recent ovulation with an accuracy of 92%. The false negative rate was reduced from 8% for a single specimen to 2.6% and 1.9% for two and three consecutive daily samples, respectively with false positive rates for a single specimen of 8% was reduced to 5.3% and 3.6% using two or three consecutive urine specimens. These findings are consistent with finding from epidemiological population-based studies that evaluated various detection algorithms based on frequencies of urine sampling to detect ovulation (34). Using daily urine sampling as the reference method, loosening this to every two or three days, but not to five day intervals, preserved accuracy of identifying recent ovulation (44) consistent with our findings of daily urine sampling frequency.

A variety of studies have pointed to the development of point-of-care or home testing for detection of ovulation (34, 45-48), but in contrast to the established use of home kits detecting the mid-cycle LH surge (with its narrow window of detection), no widely used products have emerged. One promising direction was identified by the proof that urine impregnated onto paper (urine pH strips) could be used to detect excretion of ovarian steroid glucuronides to monitor ovarian activity (13). Such filter absorbed urine steroid glucuronides were stable at room temperature for at least 6 months and avoid need for collecting and frozen storage of urine specimens. This technique proved effective for identifying ovulatory status in a population-based study of women (34). The present study demonstrates a further development suitable for home use which may prove useful for women who wish to use home monitoring, like that of self-testing for pregnancy, to plan conception, practice contraception or evaluate risk of fertility/onset infertility at perimenopause. Further medical applications of urine PDG measurements could include screening, diagnostics and therapeutics of infertility as well as investigating luteal phase defects (e.g. resulting from excessive exercise)(49).

We conclude that serial monitoring of urine PDG excretion in individual women is effective at detecting luteal phase ovulatory progesterone secretion when PDG is monitored either in urine or by DUS sampling using high specific LCMS measurement simplified to avoid need for pre-assay deconjugation or derivatization. The option of serial DUS collections stored at room temperature can simplify serial urine PDG analyses in avoiding need to store liquid urine samples making it attractive for individual’s self- monitoring of ovulatory status and in clinical research studies.

Journal Pre-proof

Page 9 of 17

References

1. Su HW, Yi YC, Wei TY, Chang TC, Cheng CM 2017 Detection of ovulation, a review of currently available methods. Bioeng Transl Med 2:238-246 2. Leiva R, Bouchard T, Boehringer H, Abulla S, Ecochard R 2015 Random serum progesterone threshold to confirm ovulation. Steroids 101:125-129 3. Thatcher SS, Donachie KM, Glasier A, Hillier SG, Baird DT 1988 The effects of clomiphene citrate on the histology of human endometrium in regularly cycling women undergoing in vitro fertilization. Fertil Steril 49:296-301 4. Roos J, Johnson S, Weddell S, Godehardt E, Schiffner J, Freundl G, Gnoth C 2015 Monitoring the menstrual cycle: Comparison of urinary and serum reproductive hormones referenced to true ovulation. Eur J Contracept Reprod Health Care 20:438-450 5. Blackwell LF, Cooke DG, Brown S 2018 The Use of Estrone-3-Glucuronide and Pregnanediol-3- Glucuronide Excretion Rates to Navigate the Continuum of Ovarian Activity. Front Public Health 6:153 6. Klopper A, Michie EA, Brown JB 1955 A method for the determination of urinary pregnanediol. J Endocrinol 12:209-219 7. Brown JB 1955 A chemical method for the determination of oestriol, oestrone and oestradiol in human urine. Biochem J 60:185-193 8. Brown JB, Blackwell LF, Billings JJ, Conway B, Cox RI, Garrett G, Holmes J, Smith MA 1987 Natural family planning. Am J Obstet Gynecol 157:1082-1089 9. Miller WL, Auchus RJ 2011 The molecular biology, biochemistry, and physiology of human steroidogenesis and its disorders. Endocr Rev 32:81-151 10. Collins WP, Collins PO, Kilpatrick MJ, Manning PA, Pike JM, Tyler JP 1979 The concentrations of urinary oestrone-3-glucuronide, LH and pregnanediol-3alpha-glucuronide as indices of ovarian function. Acta Endocrinol (Copenh) 90:336-348 11. Stanczyk FZ, Miyakawa I, Goebelsmann U 1980 Direct radioimmunoassay of urinary and pregnanediol glucuronides during the menstrual cycle. Am J Obstet Gynecol 137:443-450 12. Metcalf MG 1976 Are 24-hour urine collections necessary for the measurement of steroid excretion rates? N Z Med J 84:150-153 13. Wasalathanthri S, Tennekoon KH, Sufi S 2003 Feasibility of using paper impregnated with urine instead of liquid urine for assessing ovarian activity. Ceylon Med J 48:4-6 14. Falk RT, Gail MH, Fears TR, Rossi SC, Stanczyk F, Adlercreutz H, Kiura P, Wahala K, Donaldson JL, Vaught JB, Fillmore CM, Hoover RN, Ziegler RG 1999 Reproducibility and validity of radioimmunoassays for urinary hormones and metabolites in pre- and postmenopausal women. Cancer Epidemiol Biomarkers Prev 8:567-577 15. Cox RI 1963 Gas chromatography in the analysis of urinary pregnanediol. J Chromatogr 12:242- 245 16. Barrett SA, Brown JB 1970 An evaluation of the method of Cox for the rapid analysis of pregnanediol in urine by gas-liquid chromatography. J Endocrinol 47:471-480 17. BrookerJournal L, Cawley A, Kazlauskas R, Goebel C,Pre-proof George A 2012 Carbon isotope ratio analysis of endogenous urinary metabolites after acetate and administration for doping control. Drug Test Anal 4:951-961 18. Bowers LD, Sanaullah 1996 Direct measurement of and glucuronide conjugates with high-performance liquid chromatography-mass spectrometry. J Chromatogr B Biomed Appl 687:61-68

Page 10 of 17

19. Moneti G, Agati G, Giovannini MG, Pazzagli M, Salerno R, Messeri G, Serio M 1985 Pregnanediol-3 alpha-glucuronide measured in diluted urine by mass spectrometry with fast atom bombardment/negative-ion ionization. Clin Chem 31:46-49 20. Chen J, Hou H, Chen H, Luo Y, He Y, Wang H, Deng H, Zhang L, Zhang Y, Liu H, Liu Y, Wang A, Hu Q 2020 Quantitation of serum pregnanediol-3-glucuronide level in different endocrine statuses by LC-MS/MS. J Pharm Biomed Anal 184:113171 21. De Wilde L, Roels K, Van Renterghem P, Van Eenoo P, Deventer K 2020 Steroid profiling in urine of intact glucuronidated and sulfated steroids using liquid chromatography-mass spectrometry. J Chromatogr A 1624:461231 22. Otero-Fernández M, Cocho J, Tabernero MJ, Bermejo AM, Bermejo-Barrera P, Moreda-Piñeiro A 2013 Direct tandem mass spectrometry for the simultaneous assay of opioids, cocaine and metabolites in dried urine spots. Anal Chim Acta 784:25-32 23. Michely JA, Meyer MR, Maurer HH 2017 Dried urine spots - A novel sampling technique for comprehensive LC-MS(n) drug screening. Anal Chim Acta 982:112-121 24. Monošík R, Dragsted LO 2018 Dried urine swabs as a tool for monitoring metabolite excretion. Bioanalysis 10:1371-1381 25. Naritaka N, Suzuki M, Takei H, Chen HL, Oh SH, Kaewplang P, Zhang C, Murai T, Kurosawa T, Kimura A, Shimizu T, Nittono H 2019 Use of dried urine spots for screening of inborn errors of bile acid synthesis. Pediatr Int 61:489-494 26. Menkovic I, Marchand AS, Boutin M, Auray-Blais C 2019 Neonatal Mass Urine Screening Approach for Early Detection of Mucopolysaccharidoses by UPLC-MS/MS. Diagnostics (Basel) 9 27. Protti M, Marasca C, Cirrincione M, Sberna AE, Mandrioli R, Mercolini L 2020 Dried Urine Microsampling Coupled to Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS) for the Analysis of Unconjugated Anabolic Androgenic Steroids. Molecules 25 28. Protti M, Mandrioli R, Mercolini L 2020 Microsampling and LC-MS/MS for antidoping testing of glucocorticoids in urine. Bioanalysis 12:769-782 29. Gifford RM, Howie F, Wilson K, Johnston N, Todisco T, Crane M, Greeves JP, Skorupskaite K, Woods DR, Reynolds RM, Anderson RA 2018 Confirmation of ovulation from urinary progesterone analysis: assessment of two automated assay platforms. Scientific reports 8:17621 30. Skorupskaite K, George JT, Veldhuis JD, Anderson RA 2018 Neurokinin B Regulates Gonadotropin Secretion, Ovarian Follicle Growth, and the Timing of Ovulation in Healthy Women. J Clin Endocrinol Metab 103:95-104 31. Johnson S, Weddell S, Godbert S, Freundl G, Roos J, Gnoth C 2015 Development of the first urinary reproductive hormone ranges referenced to independently determined ovulation day. Clin Chem Lab Med 53:1099-1108 32. FDA CDER 2018 Bioanalytical Method Validation. Available at https://www.fda.gov/media/70858/download, Accessed 15 Aug 2019 33. EMEA 2011 Guideline on Bioanalytical Method Validation. Available at http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2011/08/WC50 0109686.pdf, Accessed 20 Sept 2016 34. Rosetta L, Thalabard JC, Tanniou J, Ducot B, Maitrot-Mantelet L, Rousset-Jablonski C, Bohet A, BouyerJournal J, Chimenes A, Slama R 2017 Ovulatory Pre-proof status and menstrual cycle duration assessed by self-collection of urine on pH strips in a population-based sample of French women not using hormonal contraception. Eur J Contracept Reprod Health Care 22:450-458 35. Taieb J, Mathian B, Millot F, Patricot MC, Mathieu E, Queyrel N, Lacroix I, Somma-Delpero C, Boudou P 2003 Testosterone measured by 10 immunoassays and by isotope-dilution gas chromatography-mass spectrometry in sera from 116 men, women, and children. Clin Chem 49:1381-1395

Page 11 of 17

36. Wang C, Catlin DH, Demers LM, Starcevic B, Swerdloff RS 2004 Measurement of total serum testosterone in adult men: comparison of current laboratory methods versus liquid chromatography-tandem mass spectrometry. J Clin Endocrinol Metab 89:534-543 37. Sikaris K, McLachlan RI, Kazlauskas R, de Kretser D, Holden CA, Handelsman DJ 2005 Reproductive hormone reference intervals for healthy fertile young men: evaluation of automated platform assays. J Clin Endocrinol Metab 90:5928-5936. 38. Rosner W, Auchus RJ, Azziz R, Sluss PM, Raff H 2007 Position statement: Utility, limitations, and pitfalls in measuring testosterone: an Endocrine Society position statement. J Clin Endocrinol Metab 92:405-413 39. Handelsman DJ, Desai R, Seibel MJ, Le Couteur DG, Cumming RG 2020 Circulating Sex Steroid Measurements of Men by Mass Spectrometry Are Highly Reproducible after Prolonged Frozen Storage. J Steroid Biochem Mol Biol 197:105528 40. Kesner JS, Knecht EA, Krieg EF, Jr. 1995 Stability of urinary female reproductive hormones stored under various conditions. Reprod Toxicol 9:239-244 41. Brown JB, Holmes J, Barker G 1991 Use of the Home Ovarian Monitor in pregnancy avoidance. Am J Obstet Gynecol 165:2008-2011 42. Sauer MV, Paulson RJ, Chenette P, Frederick J, Stanczyk FZ 1990 Effect of hydration on random levels of urinary pregnanediol glucuronide. Gynecol Endocrinol 4:145-149 43. Zacur H, Kaufman SC, Smith B, Westhoff C, Helbig D, Lee YJ, Gentile G 1997 Does creatinine adjustment of urinary pregnanediol glucuronide reduce or introduce measurement error? Gynecol Endocrinol 11:29-33 44. Allaway HC, Williams NI, Mallinson RJ, Koehler K, De Souza MJ 2015 Reductions in urinary collection frequency for assessment of reproductive hormones provide physiologically representative exposure and mean concentrations when compared with daily collection. Am J Hum Biol 27:358-371 45. Bouchard TP, Fehring RJ, Schneider M 2019 Pilot Evaluation of a New Urine Progesterone Test to Confirm Ovulation in Women Using a Fertility Monitor. Front Public Health 7:184 46. Leiva R, McNamara-Kilian M, Niezgoda H, Ecochard R, Bouchard T 2019 Pilot observational prospective cohort study on the use of a novel home-based urinary pregnanediol 3-glucuronide (PDG) test to confirm ovulation when used as adjunct to fertility awareness methods (FAMs) stage 1. BMJ Open 9:e028496 47. Blackwell LF, Cooke DG, Brown S 2018 Identifying ovulatory cycles and the day of ovulation by the mis-use of pregnanediol glucuronide excretion rate thresholds. Eur J Contracept Reprod Health Care 23:390-391 48. Matias-Garcia PR, Martinez-Hurtado JL, Beckley A, Schmidmayr M, Seifert-Klauss V 2018 Hormonal Smartphone Diagnostics. Methods Mol Biol 1735:505-515 49. Hakimi O, Cameron LC 2017 Effect of Exercise on Ovulation: A Systematic Review. Sports Med 47:1555-1567 Journal Pre-proof

Page 12 of 17

Tables Table 1 – Comparison of PDG measurement in urine and DUS samples

Variable Correlation1 Slope2 Intercept2 Mean Bias3 Pregnanediol glucuronide 0.98 0.99 5.0 22 (1.6%) (<0.001) (0.96, 1.03) (-8.8, 21.8) (-11, 55) Etiocholanolone glucuronide 0.98 1.02 59 -74 (5.8%) (<0.001) (0.99, 1.05) (18, 86) (-127, -20) Androsterone glucuronide 0.98 1.06 13.7 -118 (5.6%) (<0.001) (1.03, 1.08) (-15, 44) (-185, -51)

1. by Spearman rank correlation with p value in brackets 2. slope and intercept (ng/ml) by Passing-Bablok regression with 95% confidence limits in brackets 3. bias (ng/ml) by Bland-Altman deviation analysis of mean difference (urine – DUS) concentrations with mean bias in brackets and 95% confidence limits in brackets below.

Table 2 – Quantitative analysis of ROC curves

Variable PDG PDG/Cr P3G P3G/Cr P4 P4/Cr AUC 0.88 0.96 0.95 0.98 0.65 0.72 [95% CI] [0.82, [0.92, [0.90, [0.96, [0.55, [0.63, 0.92] 0.98] 0.97] 0.99] 0.73] 0.79] Optimal threshold 2372 199 3551 297 293 nmol/l 31 nmol/L µmole/mo nmol/L µmol/mol µmole/mo le e le Maximal Youden 0.64 0.86 0.79 0.88 0.23 0.36 index Minimal distance to 0.29 0.11 0.16 0.09 0.72 0.60 corner Likelihood ratio + 9.34 19.6 11.2 29.7 4.67 8.76 Likelihood ratio - 0.31 0.10 0.15 0.09 0.76 0.63 Diagnostic odds 30.6 192 76.3 329 6.15 14.0 ratio Journal Pre-proof

Page 13 of 17

Table 3 – False negative and positive rates with sequential daily urine sampling False negative False positive Direct Simulation Direct Simulation 1 sample 7.9% -- 8% -- 2 samples 2.7% 2.6% 5.8% 5.3% 3 samples 0.8% 1.9% 4.4% 3.6%

Direct represents the estimation based on enumerating false positive and negative numbers based on actual data day by day. Simulation represents a bootstrap resampling (with replacement) comprising 5000 replicates

Table 4 – Window of detection for ovulation using PDG by LCMS

PDG/CR PDG First Day 0 [-2, 3] 0 [-2, 4] Peak 5 [3, 9] 5 [1, 9] Last Day 10 [3, 13] 10 [3, 13] Window Of Detection 10 [0, 13] 8 [0, 13]

Data shown as median and [range] in days

Journal Pre-proof

Page 14 of 17

Figure Legends

Figure 1 – Plot of PDG measurement in urine versus in DUS shown as a Passing-Bablok regression (left panel) and Bland-Altman deviation analysis (right panel). For further details see text and table 1.

Figure 2 – Plot of daily urine PDG (left panel) and urine PDG adjusted for urine creatinine (right panel) over the menstrual cycle against days from 13 days before to 13 days after ovulation. Filled circles represents the mean and whiskers represent the standard error of the mean.

Journal Pre-proof

Page 15 of 17

Figure 3 – ROC curves depicting accuracy of diagnostic performance by six alternative urine hormone measurements to identify ovulation based on a single urine sample. The x axis represents the false positive rate (1-specificity) and the y axis represents true positive rate (sensitivity). The dashed diagonal line indicates the diagnostic performance of a test with no more diagnostic power than random. The red line indicates urine PDG by LCMS, the purple lines indicate P3G (PDG measured by an immunoassay) and the orange lines P4 by immunoassay. For each analyte, the dotted lines are the direct measurement and the solid lines that adjusted for urine creatinine.

Journal Pre-proof

Page 16 of 17

Journal Pre-proof

Page 17 of 17