Ultrasound and magnetic resonance imaging-based investigation of the role of perfusion and oxygen availability in menstrual pain.

There were no significant group demographic differences in age, BMI, race, ethnicity, or pelvic comorbidities in MRI participants ( Table 1 ). Whereas the participants with menstrual pain reported occasionally missed school/work, pain-free participants reported negligible impact. We assessed the effects of menses on uterine perfusion and oxygen availability within each group by comparing average R2* and correcting for multiple comparisons ( Figure 3 , Supplemental Table 1 ). Representative R2* images are shown in Figure 2 . During menses, there was a significant increase in R2* (consistent with tissue ischemia) in the myometrium (beta ± standard error of the mean: 15.74 ± 2.29 s −1 , p = 0.001, q = 0.002) and endometrium (26.37 ± 9.33 s −1 , p = 0.005, q = 0.008) in participants with dysmenorrhea. Although similar results were observed in the myometrium (28.89 ± 2.85 s −1 , p = 0.001, q = 0.002) and endometrium (75.50 ± 2.57 s −1 , p = 0.001, q = 0.003) in pain-free controls, post-hoc tests suggested menses R2* values were higher than dysmenorrhea participants (Myometrium: p = 0.008, Endometrium: p = 0.043). There were no significant changes associated with menstruation or naproxen within the rectus abdominis or vagina. We anticipated that naproxen sodium might decrease R2* (consistent with improved uterine perfusion or oxygen availability) in participants with dysmenorrhea. However, after naproxen sodium, there was an increase in R2* within the endometrium in participants with dysmenorrhea (48.29 ± 15.78 s −1 , p = 0.005, q = 0.008). Conversely, in pain-free controls, R2* decreased within the myometrium after naproxen sodium (−8.62 ± 3.25 s −1 , p = 0.008, q = 0.011). Doppler signals were analyzed from 23 participants with dysmenorrhea and 8 pain-free controls ( Table 2 ). Notably, there were no significant differences in heart rate or blood pressure between groups, either on menses or off menses, that could account for any sizeable confounding on uterine Doppler effects ( Table 2 ). We next analyzed the effects of menses on uterine artery Doppler velocity off menses and pre- and post-treatment during menses ( Figure 4 , Supplemental Table 2 )). In participants with dysmenorrhea, the mean pulsatility index (−0.42 ± 0.14, p = 0.004, q = 0.004) and the mean resistance index (−0.042 ± 0.012, p = 0.001, q = 0.001) were lower pre-treatment during menses compared with non-menses measurements. Simliar effects were observed with peak systolic velocity and minimum diastolic velocity ( Figure 4 , Supplemental Table 2 ). Pain-free control participants demonstrated an opposite, though non-significant, change ( Figure 4 , Supplemental Table 2 ). We analyzed the effects of naproxen sodium on uterine artery Doppler velocity during menses in participants with and without dysmenorrhea. Among participants with dysmenorrhea, naproxen sodium reduced menstrual pain from moderate (6) to mild (2) on 0-10 NRS ( Table 1 ). In these participants ( Figure 4 ), there was an increased pulsatility index (0.90 ± 0.24, p = 0.001, q = 0.001) and resistance index (0.050 ± 0.011, p = 0.001, q = 0.001). Conversely, in pain-free controls, there was a decrease in pulsatility index (−1.35 ± 0.22, p = 0.001, q = 0.001) and resistance index (−.106 ± 0.021, p = 0.001, q = 0.001). To determine whether effects were altered specifically during a menstrual cramp, we analyzed Doppler velocity before, during, and self-reported episodes of cramping pain ( Figure 5 ). During episodes of cramping pain on menses, there were no significant changes in Doppler parameters ( Supplemental table 3 ). Thus, changes in Doppler velocity during menses were not due to temporary effects during episodes of cramping pain. In this study, we used fMRI and Doppler ultrasonography to assess uterine physiology parameters across menstrual phases and in relation to NSAID administration during the menses. Increased R2* in the endometrium and myometrium in both participants with dysmenorrhea and pain-free controls during menses imply that uterine perfusion or oxygen availability was reduced during menstruation in both groups ( Figure 6 ). Intriguingly, during menses, participants with dysmenorrhea had lower R2* signal than pain-free controls. Also, endometrial R2* in participants with dysmenorrhea was even higher after a therapeutic naproxen dose. Given that menstrual pain participants had more uterine oxygen availability than pain-free controls--except after a therapeutic dose of naproxen sodium—impaired oxygen availability is not likely to be a primary cause for menstrual pain. The lower R2* signal during menses among participants with dysmenorrhea compared to pain-free controls, which are reversible by naproxen sodium, suggest vasodilation caused by uterine inflammation. This finding is consistent with studies that show naproxen sodium inhibits COX-2 mediated prostaglandin synthesis and prostaglandins play a critical role in inflammation. 25 Indeed, Doppler parameters suggested that uterine arterial velocity was increased among participants with dysmenorrhea during menstruation and reduced by naproxen sodium. Notably, naproxen sodium had opposite effects in pain-free controls. Thus, COX signaling may have a protective effect in pain-free controls. In any case, we found no change in Doppler parameters during spontaneous cramping pain. These results confirm that menstrual pain does not appear to be solely caused by impaired uterine perfusion or oxygen availability. Our findings confirm prior functional MRI studies that report that uterine perfusion or oxygen availability is reduced during menses and extend these findings to participants with dysmenorrhea. 7 , 26 , 27 Because menstrual pain was not evaluated in these prior studies, our study expands on the evidence that impaired perfusion and oxygen availability also occurs in dysmenorrhea, albeit perfusion and oxygen availability impairments are significantly lower. Thus, given that pain-free controls experience greater reductions in perfusion and oxygen availability, it is unlikely that sustained uterine ischemia is the primary cause of menstrual pain. This directly contradicts long-held beliefs that menstrual pain is due to uterine ischemia. 3 , 4 Conversely, recent work suggests that reduced uterine oxygenation stabilizes the molecule hypoxia-inducible factor 1 (HIF-1) and increases endometrial repair. 28 In fact, increasing uterine oxygenation in mouse models during menses decreases HIF-1α induction and delays endometrial repair. Our Doppler results are consistent with the MRI findings: increased uterine arterial Doppler velocity among participants with dysmenorrhea during menses and decreased velocity among controls could explain why R2* differs across groups. However, other Doppler studies have suggested that participants with dysmenorrhea have decreased uterine arterial Doppler velocity indices. 10 , 29 Substantial methodological differences could explain this paradox (see below). Although one may hypothesize that a uterine contraction presenting as cramping pain impairs uterine arterial perfusion—we found no evidence of this in our Doppler measurements. The uterus has a sufficient plexus of superficial branches that could evade the imposed resistance of the pressurized myometrium. 30 Additional experience is needed to establish clinical usefulness of uterine fMRI. However, our study suggests uterine fMRI can be useful for examining menstrual pain and therapeutic mechanisms. For example, in participants with dysmenorrhea and low R2*, a study of the effects of vasoconstrictive or alternative anti-inflammatory therapeutics (e.g., steroids) of pain with simultaneous fMRI could be informative. Given the paradoxical effects of naproxen sodium among pain-free controls, study of the protective effects of eicosanoids could identify novel treatments. Prior research using fMRI in the kidney has also suggested important protective effects of eicosanoids, and additional experience is expected to yield clinically meaningful tests shortly. 31 Strengths of our study include careful, targeted, and hypothesis driven MRI and ultrasound assessments, as well as a blinding of assessors to participant pain status. Unlike most prior studies investigating Doppler ultrasonography in dysmenorrhea, our offline analysis allowed for effective blinding. We also obtained measurements before and during menstruation, and specifically during cramping pain to gain the broadest understanding of the mechanisms behind menstrual pain. Our use of a squeeze bulb allowed specific analysis during cramping pain and avoided using an intrauterine pressure catheter to measure contractions that may cause confounding effects. Although our sample size was small (n=51), it is comparable to prior studies. 29 , 32 , 33 Although there are larger studies, they have other quality issues such as lack of reporting for loss to follow-up. Our statistical modeling allowed for better analysis of within-participant effects that account for missing data following principles used in pain research. 34 , 35 Unfortunately, we did not include a placebo control because we feared this would limit interest in participation. However, a recent study reported that placebo was associated with increased uterine arterial velocity 11 —which is the opposite effect of naproxen. Thus placebo effects most likely did not contribute to these results. In conclusion, our study challenges the prevailing theory of uterine ischemia as a causal factor in menstrual pain. Indeed, uterine inflammation with increased perfusion or oxygen availability could impair endometrial repair. Future studies should further investigate synthesis of prostaglandins (e.g, PGF2α, PGE), prostaglandin receptors, and COX-2 downstream mechanisms (e.g., uterine contractility, inflammation) in modulating menstrual pain symptomatology.

The NorthShore Institutional Review Board approved this study (EH14-005, 05-05-23). Participants completed a written informed consent form before study visits. Participants (ages 18-45) were recruited into an initial ultrasonography study or a subsequent fMRI study. Some participant data from these studies have already been published, 13 , 15 - 17 , but did not include Doppler ultrasonography or R2* MRI measurements—the unique focus of the current study. Participants were recruited with flyers, public transit advertisements, and physician referrals between 2015 and 2017 (see cureperiodpain.org for an example advertisement). Participants were defined as the dysmenorrhea group if they reported a moderate or severe intensity of menstrual pain equal to or greater than 6 on a 0-10 numerical rating scale (NRS; 0-no pain at all; 10-worst pain imaginable) when not taking analgesic drugs. 18 Pain-free participants were eligible if they reported no pain or limited mild menstrual cramping pain, equal to or less than 2 on a 0-10 NRS. Exclusion criteria for all participants included a history of pelvic or abdominal malignancies, infrequent menses (>45 days between periods), pregnancy within 6 months before the study, breastfeeding, body mass index >40 mg/kg 2 , and unwillingness to stop taking pain-relieving medication on the day of the study visit. Participants undergoing MRI were excluded if they had metallic implants, intrauterine devices, or claustrophobia. Among 104 potential participants that responded to advertisements, 51 total participants (n=13 pain-free controls, n=38 participants with dysmenorrhea) successfully completed study visits ( Figure 1 ). Participants in both studies completed self-administered questionnaires regarding standard demographic information and medical history in REDCap. 19 The questionnaire also covered menstrual, medication, pregnancy, and surgical history. Participants in the MRI cohort were scheduled for a study visit within the first 48 hours of menstrual bleeding onset and also during their luteal phase for non-menses sessions. Participants were instructed to abstain from acetaminophen or ibuprofen at least 8 hours before the visit or 12 hours for naproxen or opioids. We confirmed abstention of these medications and opioids before the study visit. MRI data were obtained on a 3.0T Siemens Skyra Fit scanner with an abdominal body coil. After preliminary T2 weighted anatomical sequences were performed, an R2* map (a gradient echo sequence that can be used to calculate R2*) was obtained in the sagittal plane at end-expiration during a 12-second breath-hold with multiple gradient echo sequences (FOV: 300 x 300mm, 5 slices, thickness: 5.0mm, Matrix: 256 x 256, TR: 50ms, TE: 8 equally spaced (3.09-32.3ms). 20 Following the initial scan, all participants in this cohort ingested 2 tablets of naproxen sodium (NSAID; 220mg each) with water. Approximately 90 minutes after ingestion (based on its pharmacokinetics), additional MRI sequences, including R2* maps, were obtained. Participants in the ultrasonography cohort were scheduled for a study visit within the first 48 hours of menstrual bleeding onset to acquire data during menses and also during their luteal phase for non-menses sessions. Participants were instructed to abstain from acetaminophen or ibuprofen for at least 8 hours before the visit or 12 hours for naproxen or opioids. Simultaneous with the scan, dysmenorrhea participants during menses were given a handheld rubber bulb event marker to squeeze when they experienced pain. They were instructed to squeeze the pressure bulb when they experienced increased cramping pain in proportion to the intensity of menstrual pain. Bulb squeeze data were synchronized to the ultrasound timestamp to measure the association between self-reported menstrual pain and signal intensity. 13 These squeezes defined each menstrual cramping event. Dysmenorrhea participants off menses and pain-free controls (both on and off menses) were instructed to squeeze the bulb intermittently in random 2-5 minute intervals throughout the recording session. Details regarding recording spontaneous pain in this cohort have been addressed by Kantarovich et al. 15 The uterine artery near the cervicocorporeal junction was identified with a transabdominal ultrasonography probe (2-5 MHz Convex 2D, GE Voluson 730). A 20- to 30-minute continuous recording of color Doppler velocimetry was captured while participants were instructed to squeeze the pressure bulb. Afterward, all participants in this cohort ingested 2 tablets of naproxen sodium (NSAID; 220mg each) with water. Approximately 90 minutes after administration of naproxen sodium (based on its pharmacokinetics), we repeated ultrasonography with the same bulb squeezing techniques. 21 Doppler velocity was recorded via a screen capture device and saved for offline analysis. For analysis of R2* data, image processing was performed using ImageJ (National Institutes of Health; https://imagej.nih.gov/ij/ ). Two reviewers with experience interpreting uterine MRI performed image processing blinded to the participant and group identity. Assessment of signal intensity was performed by manually tracing a freehand selection immediately around a region of interest (ROI) and quantifying via the “measure” function. The ROIs were the myometrium, endometrium, rectus abdominis, and vagina. To verify the accuracy of ROI selection, 2 reviewers analyzed the complete data set. Measurements between reviewers were strongly correlated ( r = 0.96). Correct placement of the Doppler probe was evaluated offline by a gynecologist naïve to patient categorization. Three consecutive cardiac rhythms of Doppler velocity were digitized within 15-30 seconds before a bulb squeeze (i.e., self-reported uterine pain event or random bulb squeeze), during the first 15 seconds after squeeze initiation, and 20-35 seconds after squeeze initiation. The pulsatility index, resistance index, peak systolic velocity, and minimum diastolic velocity were calculated during each time period. To verify accuracy, reviewers were trained on the same data set until they obtained a high level of repeatability on calculated parameters ( r > .95). Statistics were calculated within Stata software version 13.1 (College Station, Texas). A linear mixed regression model was used to evaluate within-subjects effects (menses before naproxen sodium, menses after naproxen sodium, and non-menses) and between-subjects effects (pain-free controls or dysmenorrhea sufferers) on either ultrasound or MRI parameters. Because some participants did not complete both menses and non-menses visits, the study was primarily powered on looking at the within-subject effect of naproxen sodium use during menses. When data are partially missing, mixed effects models are ideal. 22 Marginal effects were used to calculate the mean and standard error of the mean for each parameter. G-power (3.1) 23 was used to evaluate statistical power (α=0.05, 1-β = 0.8). Post-hoc sensitivity analyses revealed 80% power to detect a large effect size of naproxen sodium use during menses in dysmenorrhea participants for ultrasound data (Cohen’s dz = 0.66) and for MRI data (Cohen’s dz = 0.77) comparable to other related studies. 9 , 10 We controlled for multiple comparisons by using the false discovery rate method (FDR, limiting at 0.05) and calculating q values (FDR adjusted p values). 24 The data for this study are available on the open science framework.

Treatments for dysmenorrhea (menstrual pain) are often ineffective, and over half of reproductive-age patients report subsequent impairment in school, work, or social activities. 1 Differential treatment response in dysmenorrhea could be due to differences in pharmacological, anatomical (e.g., endometriosis), genetic, and neurological factors (see 2 for review). An important first step would be to verify the common hypothesis that impaired uterine perfusion or oxygenation contributes to menstrual pain 3 , 4 , and to assess how uterine hemodynamics are impacted by nonsteroidal anti-inflammatory drugs (NSAIDs). Non-invasive imaging such as ultrasound and MRI are being used for the evaluation of blood flow 5 and oxygen availability 6 in different tissues of interest. Similar to a pulse oximeter, functional MRI (fMRI) is sensitive to changes in oxygenation status of hemoglobin. Specifically the spin-spin relaxation time/rate (T2*/R2*) changes with the oxygenation status of hemoglobin.. 7 Lower T2* values or, related to this study, higher R2* values (R2* = 1/T2*) indicate relative increases in deoxyhemoglobin or decreased perfusion. 8 MRI studies reported decreased uterine T2* (suggesting increased ischemia) during menses, but they included both participants with and without dysmenorrhea. 7 , 9 Consistent with these findings, Doppler ultrasonography has suggested perfusion impairments in participants with dysmenorrhea during menses. 10 However, alleviating pain with a vasodilator, sildenafil, did not significantly affect Doppler parameters compared with placebo. 11 Thus, the role of perfusion and oxygen availability in menstrual pain remains uncertain. A modern approach must also consider the temporal profile of pain, perfusion and oxygen availability because it varies over time. Analysis of MRI studies suggest that cramping symptoms occur during uterine contractile activity. 12 , 13 Whether changes in perfusion accompany cramps also remains unclear. In one study indirectly examining contractions with intrauterine pressure monitoring, participants with dysmenorrhea had increases in uterine arterial pulsatility during these events that could indicate ischemia. 14 We performed two separate studies to assess the mechanisms underlying menstrual pain. We harnessed MRI's capability to provide operator-independent blood oxygen availability measurements to test the hypothesis that participants with dysmenorrhea have impaired uterine blood oxygen availability during menses. We also took advantage of ultrasonography’s temporal resolution and developed a new method to evaluate spontaneous pain 15 to test the hypothesis that individual cramps are associated with impaired perfusion. To fully explore these findings' importance, we assessed pain-free controls and dysmenorrhea sufferers following acute administration of naproxen sodium, a commonly used nonsteroidal anti-inflammatory drug (NSAID) for managing menstrual pain. 2