Salsalate negatively impacts microvascular function in women with endometriosis

The institutional Review Board at The Pennsylvania State University approved all experimental procedures and protocols. A Food and Drug Administration Investigational Drug Number was obtained for all protocols (IND 78,954). Verbal and written informed consent were voluntarily obtained from all participants before participation and in accordance with the guidelines set forth by the Declaration of Helsinki. Eleven women with laparoscopically diagnosed endometriosis were screened by clinical staff which included a health history questionnaire, physical examination, medical screening, and a blood chemistry analysis (Chem 24, Quest Diagnostics, Pittsburgh, PA). All participants had no history of adverse cardiovascular events and lacked cardiovascular, renal, pulmonary, neurological, or dermatological disease. Participants were not using over-the-counter or prescription medications with known primary or secondary cardiovascular effects such as antihypertensive medication, hormonal therapy, statins, or anticoagulants at the time of testing. Participants did not use tobacco products. All participants were born with a uterus and at the time of testing had at least one intact ovary. In a randomized, placebo-controlled, crossover design, participants were given 5 days of oral salsalate treatment (1500 mg b.i.d; 29 ) and five days of placebo, the final dose of either were taken on the morning of experiment visits. We allowed a minimum two-week washout period between experiments. On day 3 of oral salsalate intervention, participants underwent blood chemistry analysis to confirm plasma salicylate concentration was within the therapeutic range (10 – 30 mg/dL). Investigators responsible for analyzing data were blinded to participant treatment condition. Participants were given instructions to abstain from caffeine and alcohol for 12 hours and vigorous physical activity for 24 hours prior to experiment visits. The intradermal microdialysis technique ( 30 ) was used to assess microvascular endothelial function. Four microdialysis probes (10 mm, 55 kDa, CMA Linear 31 Probe, Harvard Apparatus, Holliston, MA) were inserted into the ventral aspect of the left forearm for local delivery of lactated Ringer’s solution (Control), 0.015 M N G -nitro-L-arginine methyl ester solution (L-NAME; Calbiochem, EMD Millipore, Billerica, MA) to inhibit NO synthase, atorvastatin (statin; 0.02 mM; USP, Rockville, MD) to act as a local anti-oxidant, and a combination of L-NAME and statin (Combo; 0.015 M L-NAME + 0.02 mM statin) dissolved in Ringer’s. A minimum of 60 minutes was allowed for resolution of the hyperemic response to microdialysis probe placement, during which time the aforementioned pharmacological agents were perfused at 2 μL/min (Hive controller and microinfusion pumps; BASi, West Lafayette, IN). A local heating apparatus (VHP2; Moor Instruments, Wilmington, DE) was placed on the skin over each inserted microdialysis probe and clamped at 33 ⁰C. Laser Doppler Flowmetry probes (VP12; Moor Instruments, Wilmington, DE) were placed in the probe holders of the local heaters to continuously measure red blood cell flux (RBC flux; perfusion units, PU). Stable baseline RBC flux was recorded for 5 – 10 minutes. Following baseline measurements, acetylcholine (ACh) was perfused in sequentially increasing concentrations (10 −10 to 10 −1 M) in combination with pharmacologic agents as described for 5-minute intervals. With ~1 min remaining in each interval, blood pressure was measured (Connex Spot Monitor, WelchAllyn, Skaneateles Falls, NY). The final 1–2 minutes of continuous RBC flux for every ACh dose was averaged and divided by mean arterial pressure to determine the cutaneous vascular conductance (CVC). Following the ACh dose response, sodium nitroprusside (28 mM; USP, Rockville, MD) was perfused and the local skin temperature was increased to 43 ⁰C to elicit maximal vasodilation ( 31 ). RBC flux was allowed to stabilize for ~5 min during this maximal phase before perfusion at each site was terminated. This stable maximal RBC flux was averaged and used to determine maximal CVC (CVC max ). During the hyperemic response to microdialysis probes insertion, flow-mediated dilation (FMD) testing was successfully performed on a subset of participants (n = 7). The participants rested in a supine position for ~10–15 minutes. An investigator (VGC) imaged the right brachial artery and measured blood flow with high-resolution ultrasonography (GE Healthcare LOGIQe Ultrasound NextGen, Wauwataosa, WI) for 1.5 min. At minimum, 1 min of stable resting values were used for analyses. A blood pressure cuff was then rapidly inflated to suprasystolic pressure (~220 mmHg; Rapid Cuff Inflation System, D.E. Hokanson Inc., Bellevue, WA), distal to the antecubital fossa for 5 min. The cuff was rapidly deflated, and the post-occlusion reactive hyperemic response was recorded for 4 min. Automated edge-detection software (Quipu Cardiovascular Suite FMD Studio, Pisa, Italy) continuously recorded the ultrasound image and blood flow throughout the protocol and measured the diameter of the artery. Movement artifact was manually removed from data analysis. To evaluate the self-reported experiences of the impact of endometriosis on quality of life, participants completed the Endometriosis Health Profile Questionnaire (EHP-30; 32 ) at each experiment visit. RBC flux was recorded at 1000 Hz and stored for offline analysis (PowerLab and LabChart, ADInstruments, Bella Vista, NSW, Australia). Two independent investigators analyzed data for both microvascular and macrovascular function, and values used for statistical analysis were averaged between them. Area under the dose-response curve (AUC) was calculated using the trapezoid rule (Prism v8.1, GraphPad Software, La Jolla, CA). CVC values at each ACh dose were normalized to site-specific maximum values (CVC %max ) for point-by-point analysis. The L-NAME-sensitive component of microvascular function was determined by calculating the difference between the AUC of the control site and L-NAME site (Control L-NAME-sensitive component) as well as the statin site and the combo site (Statin-loaded L-NAME-sensitive component). FMD was calculated as: F M D ( % ) = p e a k d i a m e t e r d u r i n g p o s t o c c l u s i v e r e a c t i v e h y p e r e m i a - b a s e l i n e d i a m e t e r b a s e l i n e d i a m e t e r × 100 Dose-response data and differences between the curves were analyzed with linear mixed effects model corrected for multiple comparisons with Tukey’s method (SAS v. 9.4; Cary NC) using mean substitutions for dose- and site-specific outliers (GraphPad Software, SanDiego, CA). Outliers were defined using the robust regression and outlier removal (ROUT) method. Non-linear regression was used for curve modeling. EC50 values were analyzed with one-way ANOVA (GraphPad Software). FMD and plasma CRP was analyzed with a paired t-test. A post hoc power analysis (G*Power 3.1 Software; α = 0.05, effect size = 0.2, power = 0.93) confirmed a sample size of n = 11 would be sufficient to detect a significant difference in microvascular and macrovascular function between placebo and salsalate intervention.

Participant characteristics are shown in Table 1 . A total of 11 women with surgically confirmed endometriosis completed the study. Nearly all were normotensive, one had high blood pressure (130/88) as categorized by the 2020 AHA guidelines ( 33 ). Three participants were on selective serotonin reuptake inhibitors, and two were utilizing thyroid hormone supplementation. Three participants were using contraception at the time of the study, though only 1 was utilizing a hormonal contraceptive pill. We confirmed participants were within safe ranges of plasma salicylate at day 3 (mean ± SD: 17.2 ± 5.5 mg/dL) of oral salsalate intervention. At day 5, all subjects were within the therapeutic range (20.7 ± 7.6 mg/dl). Plasma CRP was not different between salsalate and placebo conditions (1.9 ± 1.4 mg/dl vs 1.3 ± 0.9 mg/dl, p = 0.20). The relative CVC (CVC %max ) responses to ACh dose with and without statin are shown in Figure 1D and 1B , respectively. There was a significant interaction effect of salsalate and site (p < 0.01). L-NAME attenuated the response to ACh in the placebo (p < 0.01) but not the salsalate (p = 0.21) condition. Salsalate treatment attenuated the CVC %max response at the control site compared with placebo (p < 0.01) but did not significantly alter CVC %max at the L-NAME site (p = 0.09). The CVC %max at the combo site was not different from the CVC %max at the statin site in the placebo condition (p = 1.00) but was significantly attenuated compared with the statin site in the salsalate condition (p < 0.01). Salsalate significantly augmented CVC %max at the statin site (p < 0.01) but did not affect the response at the combo site (p = 1.00). The CVC %max response to ACh alone in the placebo condition was not different from that of the response at the statin-loaded site in the salsalate condition (p = 0.93). The response at the statin-loaded site in the placebo condition and the response to ACh alone in the salsalate condition were not different (p = 1.00). The L-NAME-sensitive portion to the ACh-induced response is shown in Figure 1E . In the placebo condition, loading the sites with statin significantly decreased the L-NAME-sensitive component of the CVC %max response (Control = 238 ± 170.5 a.u., Statin-loaded = 13.65 ± 164.5 a.u.; p < 0.01), but statin did not significantly alter the L-NAME-sensitive component in the salsalate condition (Control = 128.5 ± 111.5 a.u., Statin-loaded = 150.8 ± 169.2 a.u.; p = 0.74). Salsalate had no effect on the L-NAME-sensitive component in the control sites (p = 0.11) but increased the statin-loaded sites (p = 0.05). The curve-modeled ACh dose-responses are shown in Figure 1A and 1C . There was not a significant effect of salsalate intervention on the EC50 for the dose response curves (p = 0.19), although there was a significant main effect of site (p < 0.01). The EC50 for the L-NAME response curve was significantly greater (rightward shift) than that of the control in the placebo condition (Control, −3.9 ± 1.7 vs. L-NAME, −2.2 ± 1.7 log [Ach] (M); p = 0.03), but the EC50s were not different in the salsalate condition (Control, −3.5 ± 1.5 vs. L-NAME, −2.2 ± 1.7 log [Ach] (M); p = 0.14). The EC50 of the Combo site was not different from that of the statin site in the placebo condition (Statin −3.0 ± 1.3 vs. Combo, −2.4 ± 1.3 log [Ach] (M); p = 0.74) but was greater than the statin site in the salsalate condition (Statin −4.5 ± 1.7 vs. Combo, −2.9 ± 2.1 log [Ach] (M); p = 0.05). Absolute CVC data is shown in Table 2 . There was no effect of salsalate on baseline or max absolute CVC on any site (adjusted p = 1.00 all sites). There was no difference between the baseline or max absolute CVC of any site compared with control in either the placebo or salsalate condition (adjusted p = 1.00 all sites). All sites in both placebo and salsalate conditions demonstrated significantly higher max absolute CVC than baseline absolute CVC (adjusted p < 0.01 all sites). Flow-mediated dilation data is shown in Figure 2 . There was no difference in FMD between placebo and salsalate conditions (placebo 4.73 ± 3.89%, salsalate 4.77 ± 3.48, p = 0.79). EHP-30 results were not different between placebo and salsalate conditions (placebo 39 ± 20, salsalate 40 ± 23, p = 0.38).

Contrary to our hypothesis, five days of oral salsalate attenuated the CVC response to the endothelium-dependent vasodilator, ACh, but had no effect on the L-NAME sensitive component to vasodilation. Salsalate additionally did not affect FMD. Therefore, in young women with endometriosis, NFĸB knockdown with salsalate negatively impacted endothelium-mediated vasodilation of the cutaneous microvasculature by attenuating non-NO-dependent mechanisms. Furthermore, locally perfusing statin, a potent anti-oxidant ( 42 ), attenuated the L-NAME-sensitive component to the receptor-mediated endothelium-dependent agonist, ACh. However, locally perfusing statin after salsalate treatment seemed to counteract the attenuating effect of salsalate such that the L-NAME-sensitive component of endothelium-mediated vasodilation was not different from that of the placebo condition. Our lab has previously reported that women with endometriosis demonstrate impaired microvascular endothelial function compared with healthy women ( 24 ). The CVC response to the endothelium-dependent vasodilator, ACh, in the present sample was nearly identical to that of our previous sample. However, despite virtually identical subject characteristics, the present study shows a large NO-mediated dilation by comparison. Comparing the macrovascular endothelium function measurements of the present study to those conducted previously, we confirm that the present sample of women with endometriosis show FMD values similar or perhaps less than those reported in a similar population by other groups ( 22 , 35 ). Importantly, we have additionally shown that in a healthy cohort of young, age-, BMI-, and medication-matched women, an identical oral salsalate intervention produced a pronounced increase in the CVC response to ACh via increased non-NO-mediated vasodilation ( 36 ). Therefore, we conclude that the present cohort of women with endometriosis are demonstrating abnormal responses to established experimental methods as a result of endometriosis-specific, systemic vascular pathophysiology, both in the macrovasculature and in the microvasculature. Previous application of a similar NFĸB knockdown with salsalate in overweight/obese or aged and sedentary cohorts (i.e., systemic inflammatory conditions) significantly improved FMD ( 37 , 38 ) – a primarily NO-mediated phenomenon ( 39 ) – but had no effect on the FMD in this sample of endometriosis patients. Others have shown that doses of salsalate upwards of 4.5g/day impairs FMD in a broad range of subjects, but an effect that is primarily driven by subjects with overt CVD ( 41 ). Oral salsalate at doses within the therapeutic range attenuated endothelium-dependent vasodilation in the microvasculature via non-NO-dependent mechanisms in the present study. An alternative explanation for this result includes salsalate-induced alterations in cyclooxygenase (COX)-dependent mechanisms. ACh has been shown in healthy populations to induce vasodilation in small part through prostaglandin activity – an immediate COX metabolite – in the cutaneous microvasculature ( 34 , 40 ). While salsalate does not alter COX expression, it may affect COX activity such that expression of the downstream metabolite and potent vasoconstrictor, thromboxane, is upregulated ( 37 ). It may be that in the present study, as may be the case in other cohorts undergoing more robust salsalate intervention ( 41 ), endothelium-dependent dilation following oral salsalate is contending with thromboxane-mediated constriction. Our finding that the macrovasculature was seemingly unaffected by salsalate may actually be indicative of increased NO production counteracted by increased thromboxane production, whereas in the microvasculature, increased thromboxane production may be dominating any effect of salsalate on endothelium-mediated vasodilation. Women with endometriosis, including those in the present sample, are frequently hypercholesterolemic ( 5 ). We have previously shown that atorvastatin is successful in improving microvascular NO-mediated vasodilation in hypercholesterolemic patients ( 42 ) mediated, in part, by anti-oxidant action ( 43 ). Statin therapy in the baboon model of endometriosis, the gold standard of animal models of endometriosis, to reduce lesion load ( 44 , 45 ). We have previously demonstrated that seven days of oral statin treatment improves NO-mediated vasodilation in the microvasculature, although we were unable to delineate this finding from decreased LDL that also resulted from this short-term treatment. In this same study, we demonstrated local perfusion of atorvastatin improved endothelium-mediated vasodilation, although we did not specifically confirm the mechanism by which this occurred ( 24 ). We found in the present study that in the placebo condition, adding a local statin – and therefore applying an acute anti-oxidant treatment to the cutaneous microvasculature – abolishes the L-NAME-sensitive component to ACh-induced vasodilation. Several participants even appeared to demonstrate a reduction in vasodilation (i.e. relative vasoconstriction) in response to an endothelium-dependent vasodilatory stimulus when a local statin is administered. This collectively suggests that local reactive oxygen species (ROS) may activity contributes to NO-mediated vasodilation in women with endometriosis. We speculate that with endometriosis-associated inflammation, inducible NOS (iNOS) is active and consequently synthesizes high concentrations of NO, contributing to a large NO-mediated vasodilation that is absent when ROS activity is inhibited. There is evidence to support increased iNOS expression and NO synthesis from peritoneal fluid of women with endometriosis ( 46 , 47 ) and ectopic endometriosis lesions ( 48 ). There are a few limitations to the present study to consider. First, we did not include a healthy cohort of women in the present study to compare these findings. We have previously published findings that an identical intervention in a healthy group of women improved microvascular function ( 36 ), and therefore would not appropriately serve as a “control” group. However, we are able to examine the effect of salsalate through a placebo-controlled, investigator-blinded approach, allowing relevant examination of pathophysiologic mechanisms of vascular dysfunction in women with endometriosis. Additionally, the EHP-30 is designed to evaluate endometriosis impact on quality of life over the course of 4 weeks. Our intervention was far shorter, and so is not an adequate interrogation of the effect of an anti-inflammatory treatment on quality of life. However, this survey adds value to this study to better characterize our sample. We furthermore included three participants taking SSRIs. Like other chronic pain condition, women with endometriosis experience greater prevalence and symptom severity of mood disorders including anxiety and depression ( 49 , 50 ). Adults with major depressive disorder not treated for their symptoms have endothelial dysfunction mediated in-part by increased systemic inflammation and treatment with selective serotonin reuptake inhibitors improves endothelial function through NO-dependent mechanisms ( 51 , 52 ). In the present study, it is unclear if either the mood disorder or the SSRI treatment impacted vascular function. However, including these participants adds to the ecological validity of these findings, and removing them from the data set does not alter the significance of these findings. Finally, it is worth noting that analysis of EC50 requires a plateau of response at the higher concentrations, and this is prevented with co-administration of L-NAME. Therefore, we utilize EC50 values as a secondary evaluation tool to support the point-by-point analysis of relative CVC responses in the present study. Contrary to our hypothesis, salsalate-sensitive systemic inflammation does not induce cutaneous microvascular endothelial dysfunction nor conduit artery endothelial dysfunction in women with endometriosis. As an anti-inflammatory salsalate may have possible deleterious effects on the peripheral vasculature especially in those with elevated CVD risk. The mechanisms mediating endothelial dysfunction in the large and small vascular structures may be different within a disease context. More targeted approaches to mitigate increased CVD risk in women with endometriosis must be pursued.

Cardiovascular disease (CVD) is the leading cause of death of women ( 1 ). Endometriosis, which is estimated to impact ~10% of women, is an independent non-traditional risk factor for CVD ( 1 – 7 ). Endometriosis is a systemic inflammatory disease ( 4 , 8 – 14 ) in which endometrium-like lesions form on extrauterine sites such as the ovaries, mesentery, abdominal wall, and have even been identified in the lungs and brain ( 15 – 18 ). Systemic inflammation induces endothelial dysfunction, categorized as decreased nitric oxide (NO)-mediated vasodilation, a critical early event in the progression of overt CVD ( 19 – 21 ). Women with endometriosis demonstrate endothelial dysfunction — one of the earliest indicators of CVD risk – both in large conduit arteries ( 22 , 23 ) and in the cutaneous microvasculature ( 24 ). Seven days of systemic statin pharmacotherapy, which has potent anti-inflammatory effects ( 25 – 27 ), improves endothelial function in women with endometriosis ( 24 ). However, statins have a non-specific mechanism of action and produce a broad range of effects ( 26 , 28 ). No study to date has determined if endothelial dysfunction in women with endometriosis is directly mediated by systemic inflammation. The purpose of this study was to determine if vascular endothelial function in the micro- and macro circulations can be directly improved through an anti-inflammatory intervention (oral salsalate) in women with endometriosis. In this randomized, placebo-controlled trial, women with endometriosis were given five days of oral salsalate pharmacotherapy. This dosage regiment has been previously shown to be a systemic inhibitor of the inflammatory mediator nuclear factor ĸ B (NFĸB). We hypothesized that oral salsalate would improve cutaneous microvascular and conduit artery response to endothelium-dependent stimuli in women with endometriosis.