|Home | About | Journals | Submit | Contact Us | Français|
Many clinicians and patients believe that endometriosis-associated pain is due to the lesions. Yet causality remains an enigma, because pain symptoms attributed to endometriosis occur in women without endometriosis and because pain symptoms and severity correlate poorly with lesion characteristics. Most research and reviews focus on the lesions, not the pain. This review starts with the recognition that the experience of pain is determined by the central nervous system (CNS) and focuses on the pain symptoms.
Comprehensive searches of Pubmed, Medline and Embase were conducted for current basic and clinical research on chronic pelvic pain and endometriosis. The information was mutually interpreted by a basic scientist and a clinical researcher, both in the field of endometriosis. The goal was to develop new ways to conceptualize how endometriosis contributes to pain symptoms in the context of current treatments and the reproductive tract.
Endometriotic lesions can develop their own nerve supply, thereby creating a direct and two-way interaction between lesions and the CNS. This engagement provides a mechanism by which the dynamic and hormonally responsive nervous system is brought directly into play to produce a variety of individual differences in pain that can, in some women, become independent of the disease itself.
Major advances in improving understanding and alleviating pain in endometriosis will likely occur if the focus changes from lesions to pain. In turn, how endometriosis affects the CNS would be best examined in the context of mechanisms underlying other chronic pain conditions.
Endometriosis is an estrogen-dependent inflammatory disease that occurs in women of reproductive age, and generally becomes inactive with menopause, unless a woman uses post-menopausal hormone therapy (Goodman et al., 1989; Takayama et al., 1998; Missmer et al., 2004; Cumiskey et al., 2008).
The epidemiologic association between endometriosis and chronic pelvic pain (CPP) is suggested by the observation that among women who undergo laparoscopy, endometriosis is found in one-third who undergo surgery for CPP, compared with only 5% of those who do not have infertility or CPP (Howard, 1993, 2009). CPP often debilitates women with endometriosis for years (Sinaii et al., 2007), has a high risk of emergency department visits (Gao et al., 2006), and is associated with time lost from work and significant physical and social debility (Simoens et al., 2007). Additional expenses in women with endometriosis-related pain can arise from comorbid pain conditions like painful bladder syndrome (formerly called interstitial cystitis), migraine and irritable bowel syndrome (Mirkin et al., 2007).
Gynecologists and patients believe that CPP associated with endometriosis is caused by the endometriosis lesions (Fauconnier and Chapron, 2005). Establishing how the lesions do so has proved difficult. The spectrum of pains and lesions contributes to this challenge. Although analgesics, hormonal therapies and surgeries have been the mainstay of therapy, pain often returns, and return of pain is not necessarily associated with the return of lesions.
A fundamental concept is that the experience of pain is due to activity in the central nervous system (CNS). Thus, the relationship between pain and endometriosis could benefit by being reconsidered in the context of the nervous system. The purpose of this review, therefore, is to summarize the translational research that has recently exploded on this issue and to identify future directions.
An electronic database search (Pubmed, Medline and Embase) was performed to identify basic and clinical studies concerned with mechanisms of pain in women with endometriosis from the earliest date available, usually 1967. Mesh terms used included endometriosis, chronic pain, CPP, sensory nerve fibers, autonomic nerve fibers, sympathetic, estradiol, pelvic pain, innervation, sensitization, medical treatments including treatments by specific names, surgical treatments, including treatments by specific names and for specific sections like the reproductive tract, additional mesh terms were added. Mesh terms were used in various combinations and usually included the terms chronic pain AND endometriosis. Few studies reviewed and included in the manuscript address this relationship. Most are studies of endometriosis pathophysiology or treatment. Thus, the review is not a formal systematic literature overview, nor were quantitative analyses carried out. Instead, the authors synthesize and interpret the existing basic, clinical and translational research regarding the association between endometriosis and pain.
The constellation of pain symptoms associated with endometriosis varies from person to person. Symptoms encompass an unspecified combination of dysmenorrhea, dyspareunia and non-menstrual chronic pelvic–abdominal muscle pain with the sentinel symptom being dysmenorrhea (Table I). Non-menstrual CPP may persist for much of the month or only during specific times, such as at ovulation. Some women have additional painful symptoms such as dysuria, dyschezia and other chronic musculoskeletal conditions, which may or may not be related to endometriosis. These pain symptoms and their chronicity, patterns in relation to the menstrual cycle and association with other types of visceral pain, ultimately reflect changing actions of the nervous system.
Laparoscopic surgery has enabled surgical diagnosis of endometriosis by confirming the existence of lesions and transformed its surgical treatment (Kennedy et al., 2005). The appearance and location of endometriosis lesions, like the symptoms of pain, vary from person to person. On gross inspection, endometriosis is subdivided into three categories: superficial peritoneal endometriosis, deeply infiltrating endometriosis (DIE) and ovarian (cystic) endometriosis (Brosens et al., 1993; Koninckx et al., 1994; Nisolle and Donnez, 1997; Table I).
Visual survey at surgery is the accepted standard, but may not accurately catalogue the location, depth and extent of lesions. Furthermore, minimal and mild disease are not necessarily histologically confirmed, which may result in misclassifying cases (Marcoux et al., 1997; Vercellini et al., 2009b). Such confirmation is important for superficial peritoneal lesions, which vary to include classic (blue–black; considered ‘diagnostic’) and non-classic or subtle (clear or red, yellowish brown, white, scar-like) lesions (Martin et al., 1989; Albee et al., 2008). Recognizing variable appearance is critical because the frequency of biopsy-proven endometriosis is similar among all single-colored lesions and highest in mixed-color lesions (Stegmann et al., 2008b; Table I). Importantly, this identification only serves to confirm the diagnosis of endometriosis, not its relationship to pain.
Definitive criteria determining which, if any, endometriosis lesions cause pain symptoms are lacking. In carefully documented studies, location and extent of lesions bear little relation to location or amount of pain a woman experiences (Vercellini et al., 1996, 2007; Parazzini et al., 2001; Chapron et al., 2003; Fauconnier and Chapron, 2005). All three lesion types are associated with CPP, including those women with only minimal or mild endometriosis using the American Society for Reproductive Medicine (ASRM) classification (Fauconnier and Chapron, 2005).
Endometriomas, per se, are not associated with dysmenorrhea severity (Chopin et al., 2006), and dysmenorrhea is less frequent in women with only ovarian endometriomas compared with other locations (Vercellini et al., 1996). Endometriomas, however, may be a marker for greater severity of DIE lesions (Chapron et al., 2008).
DIE lesions are the only ones that are consistently associated with CPP (Fauconnier and Chapron, 2005). Non-menstrual CPP and gastrointestinal symptoms occur at a higher rate in those with bowel DIE lesions and painful defecation during menses with DIE involving the vagina (Fauconnier et al., 2002).
Importantly, endometriosis is associated with a local inflammatory response and formation of adhesions, be they spontaneous de novo adhesions or adhesions reformed after surgical treatment of endometriosis (Parker et al., 2005; Parazzini et al., 2006). The role of adhesions in pain and endometriosis is poorly understood (Rapkin, 1986; Parazzini et al., 2006; Vercellini et al., 2009b; Yeung et al., 2009).
Although most acknowledge that analgesic drugs alleviate pain in some women with endometriosis (Kennedy et al., 2005; Sinaii et al., 2007), conclusive evidence is lacking (Allen et al., 2009), and detailed information about their use and effectiveness in relieving pain is largely absent from gynecologic studies (Table II). Because analgesics are the first treatments used for dysmenorrhea and because these drugs are an integral part of acute and chronic pain treatment in other conditions, further research is needed on use and effectiveness of different types of analgesics in pain relief in endometriosis.
Most treatments of endometriosis suppress ovarian function, regarding endometriosis as an estrogen-dependent disease (Kennedy et al., 2005). Combined oral contraceptives (estrogen and progestin), danazol, gestrinone, medroxyprogesterone acetate and GnRH agonists are equally effective (Prentice et al., 2000a, b; Davis et al., 2007; Selak et al., 2007; Table II). During hormonal treatment, lesions, which may be quiescent (perhaps neurologically inactive), are observed in most women on GnRH agonists and progestagens (Nisolle-Pochet et al., 1988). Persistence of lesions is proposed to explain why pain returns after medical therapy is stopped. Yet, beyond their effect on lesions, most agents suppress ovulation and alter endometrium, frequently causing amenorrhea. When ovulation and menstruation are restored, often symptoms return, suggesting that these reproductive functions play a role in symptoms independent of lesions.
Osteen et al. (2005) demonstrated that eutopic endometrium of women with endometriosis has reduced progesterone action during maturation or ‘progesterone resistance’, that, in turn, may enhance development of endometriosis. The rationale for use of progestogens with or without estrogens in endometriosis treatment has been their anti-angiogenic, immunomodulatory and anti-inflammatory effects, well described by Vercellini (2003a; Table II). Progestagens and combined oral contraceptives affect eutopic endometrial cellular and paracrine characteristics such that the endometrium is fundamentally altered and becomes decidualized; these changes may inhibit implantation and growth of refluxed menstrual endometrium (Vercellini et al., 2003a).
Oral contraceptives when taken cyclically suppress ovulation and thin the endometrium (Grow and Iromloo, 2006), resulting in lighter menses. No one type of oral contraceptive is more effective than another, nor is there evidence that taking oral contraceptives continuously without placebo pills is superior in relieving pain to cyclic oral, percutaneous (contraceptive patch) or transvaginal (contraceptive ring) regimens (Vercellini et al., 2003c, 2009a; Table II).
Oral contraceptives and progestagens are the most recommended treatment for endometriosis because of their good safety profile, low cost and tolerability (Kennedy et al., 2005). When these agents are ineffective, poorly tolerated or contraindicated, GnRH analogues, danazol or gestrinone are used, which cost more and are less well tolerated.
GnRH agonists are sometimes used prior to laparoscopic diagnosis; if symptoms are relieved, some presume that the woman has endometriosis. Leuprolide acetate, for example, was effective for pain in one randomized clinical trial; importantly, women without endometriosis also had pain relief (Ling, 1999), a finding confirmed by others (Jenkins et al., 2008). Treatment with GnRH agonists can be safely administered for up to 2 years when combined with add-back hormonal therapy for bone protection (Surrey, 1999; Surrey and Hornstein, 2002), although concerns remain about use in women who have not reached peak bone mass (Divasta et al., 2007; Laufer, 2008).
The levonorgestrel intrauterine system (IUD) may also be effective in improving pain symptoms (Vercellini et al., 1999, 2003b; Lockhat et al., 2004, 2005; Petta et al., 2005; Table II), but there is insufficient evidence to recommend its use (Abou-Setta et al., 2006; Varma et al., 2006). It causes decidualization of endometrial stroma, endometrial glandular atrophy with down-regulation of endometrial cell proliferation and increase in apoptotic activity (Vercellini et al., 2005).
Antiprogestins may be effective in endometriosis treatment (Kettel et al., 1998; Prentice et al., 2000a), but have not had widespread use (Table II). Similarly, aromatase inhibitors have been effective in non-controlled studies of women who have pain unresponsive to other hormone treatments (Bulun et al., 1999; Amsterdam et al., 2005). However, use of aromatase inhibitors requires ovarian suppression with another hormonal agent (Table II).
A lack of relationship between lesions and CPP, and ineffectiveness of raloxifene in relieving pain symptoms were shown in a randomized, double-blind, placebo-controlled study of raloxifene taken after endometriosis excision. Unexpectedly, raloxifene significantly shortened the time to return of pain (Stratton et al., 2008; Table II). At second laparoscopy, many women in both groups had biopsy-proven endometriosis; those taking raloxifene underwent second surgery significantly sooner, and raloxifene use, but not endometriosis, was associated with return of pain. Because recurrence of endometriosis lesions was not associated with return of pain, estrogen or factors other than the lesions per se, likely caused their pelvic pain.
The higher levels of peritoneal macrophages in endometriosis compared with healthy women (Weinberg et al., 1991) and the finding that endometriosis activates prostaglandins to perpetuate local inflammation (Taylor et al., 1997; Bulun, 2009) suggest that local inflammation is a critical contributor to the relationship between pain and endometriosis. Thus, one new focus of treatments involves agents that can inhibit immunological and inflammatory factors, including angiogenesis inhibitors (Becker and D'Amato, 2007; May and Becker, 2008), tumor necrosis factor-alpha (TNF-alpha; Barrier et al., 2004; Koninckx et al., 2008) and peroxisome proliferators-activated receptors gamma (Pritts et al., 2002; Tee et al., 2006). Of those studied in clinical trials, TNF-alpha antagonists have not been shown to be effective (Koninckx et al., 2008; Lv et al., 2010). Chinese herbal therapy appears also to inhibit immunological and inflammatory factors in vitro (Wieser et al., 2007, 2009) and has shown some efficacy in clinical trials for endometriosis-associated pain (Flower et al., 2009).
The approach to surgical treatment of endometriosis in women with CPP is based on the oncologic principle of removing all lesions (Howard, 2003; Vercellini et al., 2009b). Yet, pain recurrence and re-operation rates are very high, with re-operation rates for symptoms of 50–60% by 5–7 years (Cheong et al., 2008; Shakiba et al., 2008). In the 19–29 year age group, over 70% have another surgery. Shakiba et al. (2008) reported that re-operation rates for symptoms are lower after hysterectomy than after operative laparoscopy, a finding also reported by Endometriosis Association members (Sinaii et al., 2007), suggesting the uterus itself contributes to pain in these women. Furthermore, not all women undergoing re-operation have endometriosis at second surgery (Vercellini et al., 2009b), which likely reflects independent central neural contributions to pain or that pain is due to something other than endometriosis.
Only three published randomized clinical studies evaluate the effectiveness of surgery (Sutton et al., 1994, 1997; Jones et al., 2001; Abbott et al., 2004; Jarrell et al., 2005; Table II). In these well-designed studies, limited sample size (131 women total), brief follow-up, and loss to follow-up severely hampered evaluating long-term relief of symptoms after surgery. In contrast, large numbers of published observational and retrospective studies illustrate that some surgeons publish good experience and that surgery is operator-dependent.
The variable appearance and location of lesions renders surgical diagnosis of endometriosis and treatment of lesions complex (Table II). These issues contribute to the dilemma of determining which lesions contribute to pain. For example, one clinical study (for infertility) accepted only bluish-black lesions as documentation of endometriosis (Marcoux et al., 1997). The variability across studies in lesion criteria and whether all visible, suspicious lesions were identified and treated further compounds the difficulty in understanding which lesions cause pain.
An important yet unstudied issue is the surgical technique used to treat lesions (Table II). Superficial lesions may be ablated (destroyed in place) or excised (removed). Whereas ablating lesions is technically easier, ablation may promote adhesions. The charred tissue itself can be interpreted as endometriosis at subsequent surgery, especially if not biopsied. Prospective studies comparing techniques have not been conducted as each surgeon chooses the technique they prefer and their skill by technique may vary.
Treating endometriomas is not straightforward as their association with pain is unclear (Chapron et al., 2002; Vercellini et al., 2009b). Although excision of the pseudocapsule helps to prevent recurrence, excision may lead to diminished ovarian reserve, subsequent infertility and premature ovarian insufficiency (Vercellini et al., 2006b, 2009b).
Treating DIE lesions has a greater potential for benefit than treating other lesion types (Fauconnier and Chapron, 2005; Vercellini et al., 2009b). However, DIE surgeries have more complications (3–10%), including blood loss, hemoperitoneum, fistula formation, pelvic abscess and bladder denervation. These complications underscore that DIE treatment requires advanced surgical skill and may require assistance of other specialists including gastrointestinal surgeons, pelvic/reconstructive surgeons, urologists or urogynecologists.
Timing of surgery during the menstrual cycle may be important in preventing recurrent disease, adhesions and symptoms (Schweppe and Ring, 2002; Table II). As peritoneal healing takes 3–5 days, if retrograde menstruation occurs on healing peritoneum, endometriosis and adhesions may immediately reform. This scenario parallels an in vitro model for endometriosis used by Schenken (Witz et al., 2002). Thus, some surgeons favor operating only in the woman's follicular phase after menses (Koh and Janik, 2002; Schweppe and Ring, 2002), but most studies do not report menstrual phase at surgery.
A complicating factor is that adhesions, lysed at surgery, form again (Parker et al., 2005; Table II). Sites of adhesions with or without endometriosis at index surgery are more likely to reform adhesions than sites of endometriosis lesions alone. Thick adhesions are more likely to reform than thin ones. Yet, whether these adhesions contribute to return of pain symptoms is unknown.
Studies on the effectiveness of surgery for symptomatic endometriosis illustrate that 20% of patients do not have relief after surgery and those who fail are more likely to have minimal or mild endometriosis (Sutton et al., 1994; Abbott et al., 2004; Jarrell et al., 2005). Furthermore, pain symptoms return sooner in women with minimal or mild endometriosis than those with moderate or severe disease (Sutton et al., 1994). Others report that stage of disease is not associated with differences in time to repeat surgical treatment (Jarrell et al., 2007).
Furthermore, in case series, women with rectovaginal disease also experience lesion recurrence (Fedele et al., 2004; Brouwer and Woods, 2007; Kristensen and Kjer, 2007) and 25% undergo repetitive conservative or definitive surgery to treat pain relapse (Reich et al., 1991; Nezhat et al., 1992; Mohr et al., 2005; Mereu et al., 2007).
Finally, other surgical strategies have failed in women with endometriosis and pain. Redwine removed all pelvic peritoneum, yet symptoms and some endometriosis returned (Redwine, 1997; Redwine and Wright, 2001). Endometriosis-associated dysmenorrhea treated by either laparoscopic uterine nerve ablation or presacral neurectomy had mixed results (Gambone et al., 2002; Proctor et al., 2005; Daniels et al., 2009). Similarly, locating and excising painful peritoneal areas during laparoscopy by conscious pain mapping had limited success (Howard et al., 2000; Steege, 2001).
Three types of treatment alleviate pain in women with endometriosis: non-steroidal (NSAID) and other analgesics, hormonal agents and surgical treatment of the lesions. The fact that NSAIDs alleviate pain in some women suggests that endometriosis increases pro-inflammatory agents that, in turn, contribute to pain. Although much is known about how these agents affect lesions (Kyama et al., 2009), little is known about how pro-inflammatory factors contribute to pain symptoms other than the consensus that their release into peritoneal fluid could activate nearby sensory fibers, thereby eliciting pain (Bulun, 2009). Furthermore, other factors are likely involved because NSAIDs are often ineffective (Allen et al., 2009), even when combined with surgical removal of lesions or hormonal therapy (Kennedy et al., 2005; Appleyard et al., 2006; Ozawa et al., 2006; Practice Committee of the American Society for Reproductive and Medicine, 2008; Vercellini et al., 2009b)
The fact that hormonal treatments that reduce estradiol levels or influence progesterone alleviate pain in women with endometriosis supports the conclusion that estradiol contributes to pain symptoms and that progestogens may contribute to decreased symptoms (Fedele et al., 2008; Huang, 2008; Aghajanova et al., 2009). Again, however, how hormones exert their effects on pain in endometriosis is poorly understood. For example, whereas ‘progesterone resistance’ may be important in the development of endometriosis, whether endometriosis-associated pain stems from ‘progesterone resistance’ is unknown. Furthermore, other factors are involved, because hormonal treatments do not alleviate pain in all women with endometriosis (Vercellini et al., 2003a; Attar and Bulun, 2006; Jenkins et al., 2008), yet leuprolide relieves pain in women, regardless of whether they have endometriosis (Ling, 1999, Jenkins et al., 2008).
The fact that surgical removal or destruction of lesions alleviates pain in some patients indicates that lesions contribute to pain (Vercellini et al., 2009b), but how they do so is poorly understood. First, seemingly complete surgical removal fails to alleviate pain for at least a year in up to 50% of carefully selected patients (Vercellini et al., 2009b). Second, even in patients whose pain was alleviated by surgically treating lesions, pain often returns, sometimes without evidence of new lesions (Abbott et al., 2003; Vercellini et al., 2009b). Third, severity of pain symptoms does not correlate with extent of disease (Kennedy et al., 2005; Vercellini et al., 2009b). Fourth, patients with the least amount of disease, appear more likely to re-experience pain soon after surgery (Sutton et al., 1994). Thus, pain symptoms experienced by those with few lesions may reflect a remodeling of the CNS that is not affected by removing those lesions. Fifth, in the absence of tissue diagnosis, especially with minimal or mild disease, it is unknown whether these women have endometriosis and, thus, whether their potential lesions contribute to their pain. Sixth, as surgical treatment is likely operator-dependent, not only may some surgeons be more skilled in completely removing lesions, but skill in recognizing variable lesion appearance may differ. Seventh, while some lesions are associated with pain more than others, such as DIE versus ovarian endometriomas (Koninckx et al., 1991; Fauconnier and Chapron, 2005), the basis for this difference is not yet understood.
Research so far encourages consideration of endometriosis as a chronic inflammatory disease, which unfortunately means that in most cases a single laparoscopic procedure or a 3–6-month medication prescription will not be sufficient. Currently, we cannot predict which women will experience lasting relief of pain symptoms by surgical removal of lesions. Nor can we predict which women will be relieved by medications taken long-term, like oral contraceptive pills. As yet unstudied is whether those with worsening pain with hormonal treatment have progesterone resistance (Ryan and Taylor, 1997; Osteen et al., 2005). Novel treatments like Chinese herbal medications that inhibit angiogenesis and inflammation observed with endometriosis need more study and whether their potential efficacy is related to neural effects is discussed below. Similar to other pain conditions, endometriosis encourages a multi-therapeutic approach to its treatment (Table III).
In order to understand how endometriosis is associated with pain, it is worthwhile to begin with first principles: pain for any individual is created by activity in that individual's CNS. Thus, the question becomes, how and under what circumstances does endometriosis engage the CNS to evoke different pain symptoms (Lundeberg and Lund, 2008; Bulun, 2009; Wang et al., 2009)?
Few hypotheses address how lesions engage the CNS to produce pain. Lesions may produce pain by compressing or infiltrating nerves near lesions (Woolf, 1996; Ramer and Bisby, 1999; Anaf et al., 2000). The presence of nerve growth factor (NGF) in lesions may be involved, especially in deep adenomyotic nodules where it correlates with hyperalgesia, defined as intense pain reported when pressure is exerted in the posterior fornix (Anaf et al., 2002). NGF can also act on those nerves to produce pain and facilitate nerve growth (Mendell et al., 1999; Cheng and Ji, 2008). Tamburro et al. (2003) report an association between dysmenorrhea severity and transforming growth factor beta1 (TGFβ1) in nerve fiber bundles adjacent to endometriotic lesions; more TGFβ1 is found in lesions that are either DIE or red compared with black lesions or healthy peritoneum.
The concept of surrounding or compressing nearby nerves does not explain pain in situations where nerves are not near lesions. Indeed, pelvic pain is not correlated with nerve fibers in adhesions, peritoneum or within endometriomas (visualized by antibodies to neurofilament protein (Tulandi et al., 1998; Tulandi et al., 2001; Al-Fozan et al., 2004). Importantly, however, these studies call attention to ‘nerves’.
Endometriosis lesions must be vascularized to attach and survive (May and Becker, 2008). Importantly, blood vessels are innervated by sensory and sympathetic fibers (Burnstock, 2009) such that factors that act on sprouting blood vessels also act on nerve fibers [vascular endothelial growth factor (VEGF), NGF, semaphorins, netrins, slits and membrane-bound ephrins; Carmeliet and Tessier-Lavigne, 2005; Jones and Li, 2007; Raab and Plate, 2007). Thus, when blood vessels branch to vascularize developing lesions (‘angiogenesis’), nerves innervating those blood vessels may also branch (‘neural sprouting’) thereby enabling nerves to invade lesions.
Direct innervation of ectopic endometrial growths by sensory and sympathetic fibers was shown in endometriosis growths from a rat model (Berkley et al., 2004) and lesions from women (Berkley et al., 2005). Finding that lesions can be directly innervated by newly sprouted fibers gave rise to a new conceptualization of mechanisms underlying pain in endometriosis: characteristics of the ectopic growths' newly formed nerve supply and resultant two-way communication between the growths and CNS contribute to the variable association between the growths and pain (Berkley et al., 2005). Such sprouting, especially if associated with NGF likely sensitizes sensory fibers (Mendell et al., 1999), meaning that their spontaneous activity is greater or they are more easily activated by stimulation.
In clinical studies, using immunohistochemical techniques to mark various neuronal fiber types, Fraser's group demonstrated sensory, sympathetic and parasympathetic fibers in different types of lesions, reporting that innervation is denser in DIE than other lesion types, and densest in rectal DIE lesions (Tokushige et al., 2006b; Fraser et al., 2008; Wang et al., 2009). Importantly, Mechsner et al. (2009) found that nerve fiber density correlated with severity of pelvic pain or dysmenorrhea.
Mechanisms by which this new nerve supply could contribute to pain are being studied in a rat model (Berkley et al., 2004). The model involves autotransplanting pieces of uterine horn (ENDO), or fat (shamENDO) on abdominal arteries (Vernon and Wilson, 1985). The uterine, but not fat transplants become vascularized and grow rapidly in the first month, stabilize by 2 months, and remain viable >10 months (Vernon and Wilson, 1985). The ENDO model displays pain symptoms like those occurring in women with endometriosis. ENDO rats develop vaginal hyperalgesia (Cason et al., 2003), which is accompanied by increased abdominal muscle activity (Nagabukuro and Berkley, 2007b). As in women, symptom severity in ENDO rats does not correlate with the volume of ectopic growths (Nagabukuro and Berkley, 2007b).
Table IV lists currently known similarities and differences between this rat model and women with endometriosis. Despite the differences, studies using the model provide clues for how the ectopic growths' nerve supply could contribute to pain in women via peripheral and central neural factors (Fig. 1).
In the ENDO model, vaginal hyperalgesia severity varies with the ovarian cycle and is paralleled by changes in ectopic cysts (Zhang et al., 2008). As severity decreases, the cysts' sympathetic innervation, NGF and VEGF significantly decrease, along with increases in the cysts' vascularization. Neural changes do not occur in the rat's eutopic uterus. In addition, both sympathetic and sensory fibers in cysts co-label with antibodies to the NGF receptor (TrkA), indicating that NGF can act on both types (Sofroniew et al., 2001). The findings imply that cysts are directly conveying hormonally modifiable information to the CNS while simultaneously receiving hormonally modifiable information from the CNS (Fig. 1, Part 1). Similar functional interlinking between sympathetic and sensory innervation is found in other animal and human neuropathic and chronic pain conditions (Janig et al., 1996). Zhang et al. (2008) suggested that endometriosis is a ‘neurovascular condition’ (much like headache), a concept that was also suggested clinically (Giamberardino et al., 2001; Bendtsen, 2003; Calandre et al., 2006; Evans et al., 2007). The rat data suggest that, within ectopic growths, at least four features contribute to severity of endometriosis pain: (i) systemic and local hormonal environment; (ii) local growth factor environment; (iii) vascularity; and (iv) cysts’ sensory and sympathetic innervation.
Sympathetic nerve involvement and peripheral estradiol modulation are being studied in women. Possover et al. (2009) found a strong association between sympathetic fibers and blood vessels in endometriosis lesions infiltrating uterosacral ligaments. However, correlation with patients’ symptoms was not assessed. Signorile et al. (2009) reported a significant correlation between the immunohistochemical presence of estrogen receptor alpha, progesterone receptors, nerve fibers and vascularity in rectovaginal DIE lesions. Although immunohistochemical parameters did not correlate with scores of disease, symptoms or contraceptive steroid use, 92% of women had CPP. Stratton et al. (2008) found an association between raloxifene use and accelerated return of pain after surgery. Because raloxifene use likely increases circulating estradiol levels (Baker et al., 1998), these findings support the potential role of peripheral or central estradiol modulation.
Sensory fibers whose branches innervate ectopic growths immunostain with antibodies to calcitonin gene-related peptide (CGRP) (Berkley et al., 2004, 2005), indicating that the CGRP-positive fibers include C-fiber nociceptors (Snider and McMahon, 1998). Nociceptors are peripheral sensory detectors that respond to a noxious stimulus, defined as one that is potentially or actually injurious (Merskey and Bogduk, 1994; Sherrington, 1906). Many C-fiber nociceptors are normally ‘silent’ (Cervero and Janig, 1992; Michaelis et al., 1996); when activated by noxious events such as inflammation, not only do they convey information to the CNS (afferent function), but they also can release substance P, CGRP, tachykinins, somatostatin, nitric oxide and other factors into the local environment (efferent function). Once activated, C-fibers can become sensitized, meaning that they no longer remain silent even after inflammation resolves (Gebhart, 1999; Fig. 1, Part 1). Efferent actions also increase local vascular permeability and inflammation, a process called ‘neurogenic inflammation’ (Holzer, 1998, Szolcsanyi, 2004).
Sensory fibers innervating ENDO rats’ ectopic growths are derived from pre-existing nerve fibers that innervate adjacent regions (Zhang et al., 2008). This branching of CGRP-positive sensory fibers strongly suggests that they have become sensitized (Janig et al., 1996), meaning they exhibit ongoing electrical activity (Dmitrieva et al., 2009).
Findings from the rat model strongly implicate CNS contributions and their modulation by estradiol, likely triggered by sensitized innervation of the growths (Fig. 1, Part 2). First, effects of ENDO on nociception (Table IV) occur in pelvic/abdominal regions (bladder, vagina, abdominal muscle and mid-ureter). Afferent fibers from these regions enter spinal segments at a different level from neural input from cysts. This anatomical separation suggests that intersegmental neural processing occurs (Fig. 1, Part 3). Second, ENDO's effects on abdominal muscle activity and vaginal hyperalgesia are significantly greater when estradiol is high than when low. These effects may reflect estradiol modulation of neuronal processing within the CNS (Evrard and Balthazart, 2003; Ji et al., 2003b, 2005; Evrard and Balthazart, 2004; Tang et al., 2008) rather than estradiol's direct effects on cysts or peritoneal environment (Cason et al., 2003; Nagabukuro and Berkley, 2007b). Third, completely removing the cysts abolishes ENDO-induced vaginal hyperalgesia, but returns when rats undergo reproductive senescence (McAllister et al., 2009) at which time their estradiol levels, unlike women and mice, increase significantly (Lu et al., 1979; Berkley et al., 2007). Thus, the CNS can retain a ‘memory’ of central neuronal changes induced by the cysts’ neural input that can be ‘recalled’ by estradiol actions on activity of CNS neurons. Fourth, ENDO-induced vaginal hyperalgesia increases when cysts are incompletely removed (McAllister et al., 2009). Surgery may exacerbate the activity of peripheral afferent fibers remaining in cysts. However, because increased vaginal hyperalgesia occurs in this situation even when estradiol levels are low, CNS neurons may be released from estradiol modulation in cycling rats. Fifth, ENDO, but not shamENDO, increases activation by bladder distention of spinal neurons located in the L6/S1 spinal segments (McGinty et al., 2009), directly supporting central involvement in vaginal hyperalgesia.
These rat model findings suggest that central sensitization, a process thought to underlie pain hypersensitivity, is an important mechanism contributing, along with peripheral sensitization, to the association between endometriosis and pain (Woolf, 1983; McMahon, 1991; Coderre et al., 1993; Giamberardino et al., 1997; Ren and Dubner, 1999; Melzack et al., 2001; Ji et al., 2003b; DeLeo, 2006; Woolf and Ma, 2007; Ren and Dubner, 2008; Fig. 1, Parts 2–5, asterisks). Central sensitization is defined as ‘an increase in the excitability of the CNS so that normal inputs now evoke exaggerated responses’ (Woolf, 1983). Clinical evidence suggests this concept applies to painful endometriosis in women (Bajaj et al., 2003).
Central sensitization is initiated by peripheral sensitization (Fig. 1) and maintained by continued input to the CNS from sensitized sensory afferent fibers (Woolf and Salter, 2000). Changes in activity of peripherally sensitized fibers, due to estradiol and sympathetic modulation in the rat model, can ‘modulate’ activity of centrally sensitized neurons to affect pain severity (Woolf and Salter, 2000). Elimination of sensitized input, such as removal of ectopic growths, would relieve pain.
Sometimes, however, central neural processing can be ‘modified’ so that initially peripherally induced central sensitization becomes independent of peripheral sensitization (Woolf and Salter, 2000) via neural mechanisms similar to those underlying memory (Melzack et al., 2001; Ji et al., 2003a; Ren and Dubner, 2007). The pain then remains long after the initiating pathophysiology resolves. As a result, therapies directed at the periphery, including removal of ectopic growths or medical therapy to alter them, fail to relieve pain, and pain becomes difficult to treat.
Finding other pain syndromes such as painful bladder or irritable bowel syndrome in women with endometriosis-associated pain suggests that mechanisms underlying cross-talk between viscera contribute to pain, as they do in the rat model (Giamberardino et al., 2002; Morrison et al., 2006). Li et al. (2008), provide evidence in rodents that uterine inflammation can directly affect the colon via increased phosphorylated extracellular signal-regulated kinase (pERK) and substance P in peripheral sensory neurons (dorsal root ganglion, DRG, neurons) that branch to supply both organs. However, most DRG neurons supply one organ, rarely branching to supply more than one (Winnard et al., 2006; Li et al., 2008). Therefore, most cross-organ effects likely arise from convergence of input from different organs on spinal cord neurons.
Estradiol can influence sensitivity of peripheral sensory neurons as well as sympathetic nerve actions on those neurons (Komisaruk et al., 1972; Kow and Pfaff, 1973; Adler et al., 1977; Kaur et al., 2007). Further, evidence suggests estradiol modulates central neuronal activity in pain states (Craft et al., 2004; Greenspan et al., 2007). For example, alterations in circulating estradiol produce complex alterations in how neurons in the thalamus process information concerning the reproductive organs, colon and skin (Reed et al., 2009).
Information processing by the highly interconnected CNS circuitry can be significantly influenced by peripheral sensitization in endometriosis (Fig. 1). Thus, sensitized inputs from ectopic growths can influence neuronal activity throughout the CNS by unmasking latent, or by altering existing excitatory or inhibitory processes. Some central effects are maintained by continued input from the ectopic growth's sensory fibers, which themselves are modulated by estradiol and activity in sympathetic fibers. Production of local peritoneal factors by implants themselves, or during menstruation or ovulation may also contribute to maintaining continued input to the CNS. Further, other central actions can become independent of any peripheral input from ectopic growths due to long-term ‘modification’ of CNS functioning. All central effects can be modulated by estradiol.
Most central sensitization occurs at the spinal entry region of sensitized peripheral afferent fibers (Fig. 1, Part 2). Thus, in women, pain occurs mostly in the pelvis because most lesions are located there (Vercellini et al., 2007; Bricou et al., 2008). Pain would involve not only direct association with innervated-lesion-endowed pelvic reproductive organs, but also pains referred to abdomino-pelvic muscles, bladder and colon. Importantly, ‘remote’ central sensitization can also occur via long-ranging peripheral sensory fibers from disparate parts of the body and enormous CNS interconnectivity (Fig. 1, Parts 3–5). For women, remote central sensitization could account for co-morbidity of pelvic pain with other pain syndromes such as temporomandibular joint disorder or fibromyalgia (Sinaii et al., 2002; DeSantana and Sluka, 2008). Further, estradiol-dependence of endometriosis's pain symptoms can be partly explained by estradiol's effects on central neuronal functioning.
Pain researchers are beginning to study treatments based on the presumed mechanism of the patient's pain rather than disease associated with pain (Woolf, 2004), for example, neuropathic pain (Dworkin et al., 2010). Endometriosis has not yet been widely recognized as a disease associated with neural dysfunction such as central sensitization, and thus therapies directed at the nervous system have not yet been studied in clinical trials of endometriosis-associated pain. However, endometriosis researchers are beginning to focus on novel compounds that ultimately may have neural affects via growth factors, inflammation, angiogenesis inhibitors and the endocannabinoid system (Becker and D'Amato, 2007; Nagabukuro and Berkley, 2007a; Oktem et al., 2007; Zhang et al., 2008; Dmitrieva et al., in press).
Furthermore, because hormonal and surgical therapies are too often unsuccessful in the long-term, women and clinicians have in fact resorted to multi-therapeutic strategies commonly used in management of chronic pain (Schatman and Campbell, 2007;Berkley and Stratton, 2009; Table III).
Altered cytokine production resulting in elevated pro-inflammatory cytokines is a key aspect of chronic pain states. Chronic pain in animal models is associated with increased pro-inflammatory cytokines in systemic circulation (Walters, 1994; Watkins et al., 1995a, b; Wieseler-Frank et al., 2005). Administration of inflammatory cytokines TNFα, interleukin-1 (IL-1) and IL-6 in animal models produces both the illness response and increased sensitivity to noxious stimuli (Berczi et al., 1998; Coelho et al., 2000; Wieseler-Frank et al., 2005). Lipopolysaccharide administered peripherally in rodents results in enhanced pro-inflammatory cytokine expression and marked hyperalgesia that is abolished by administration of an IL-1 receptor antagonist (Maier et al., 1993; Taylor et al., 1998; Maier and Watkins, 2003) or TNFα antagonist (Coelho et al., 2000; Safieh-Garabedian et al., 2000). These same inflammatory cytokines are elevated in peritoneal fluid (Koninckx et al., 1998; Minici et al., 2007) in women with painful endometriosis and such factors are an important facet of central sensitization (Wieseler-Frank et al., 2005). However, these treatments have not been effective for endometriosis-associated pain (Koninckx et al., 2008).
Pain is a stressful experience, and animal models show that the hypothalamic–pituitary–adrenal (HPA) axis is activated during both tonic and phasic pain (Taylor et al., 1998). Monoaminergic neurons in the brain stem usually act as brakes to nociceptive transmission. However, with HPA activation, an increase in glucocorticoid secretion induces monoamine depletion and loss of monoaminergic tone (Blackburn-Munro and Blackburn-Munro, 2003), resulting in increased pain.
Chronic stress is associated with enhanced pain (Romero and Sapolsky, 1996). The continuous and prolonged activation of the HPA axis causes the negative feedback loop to become dysregulated, leading to either enhanced hormone production and release or resistance to circulating glucocorticoids (Blackburn-Munro and Blackburn-Munro, 2003). Additionally, in chronic stress, loss of glucocorticoid receptor expression can cause a loss of glucocorticoid inhibition of pro-inflammatory cytokines that results in an increase in cytokine levels (Sapolsky et al., 2000). The resulting chain of events can contribute to peripheral sensitization and, eventually, central sensitization.
Clinical studies on chronic pain states such as fibromyalgia, rheumatoid arthritis and several autoimmune inflammatory diseases implicate involvement of dysfunctional HPA responsivity (Buyalos et al., 1992; Lentjes et al., 1997; Griep et al., 1998; Bellometti and Galzigna, 1999; Torpy et al., 2000; Crofford, 2002). The same situation likely applies to endometriosis and chronic pain (Stegmann et al., 2008a).
Given that peripheral and central sensitization contribute to pain symptoms, the question then arises as to how reproductive system functioning also contributes. Peripheral factors associated with painful endometriosis are found in women without endometriosis, such as increased eutopic uterine innervation (Quinn and Kirk, 2002; Tokushige et al., 2007; Zhang et al., 2009) and high concentrations of prostaglandins and other pro-inflammatory agents in peritoneal fluid (Koninckx et al., 1998). Dynamic events that repeat monthly during a woman's menstrual cycle (ovulation, endometrial maturation and shedding) and change over her reproductive life are intertwined. Menstrual cycle and other reproductive events, like pregnancy, likely influence sensitization and pain in women with endometriosis.
Throughout the menstrual cycle, the endometrium produces cytokines and growth factors in response to variations in ovarian steroids (Critchley et al., 2006; Jabbour et al., 2006; Collins and Crosignani, 2007; Table V). With withdrawal of progesterone during the premenstrual phase, stroma, endothelium and immune cells undergo changes, inducing local mediators such as chemokines, cytokines, cell regulators and enzymes responsible for prostaglandin synthesis and degradation. These mediators and enzymes, in turn, facilitate influx of leukocytes into tissues (Critchley et al., 2001). VEGF and its receptor VEGFR-2 kinase insert domain receptor (KDR) are markedly up-regulated in superficial endometrial stromal cells in human and non-human primate models (Nayak et al., 2000; Brenner et al., 2002) In all women, menstrual endometrium is rich in leukocytes, chemokines, cytokines, VEGF, KDR, matrix metalloproteinases (MMPs) and prostaglandins (Salamonsen et al., 2002), some of which, if altered, likely contribute to nerve sprouting and perpetuating sensitization. Hormonal treatments may diminish these factors (see review by Vercellini et al., 2003a)
In endometriosis, eutopic endometrium exhibits several subtle but possibly important molecular changes, including cascades favoring increased synthesis of cytokines, prostaglandins, metalloproteinases and estrogen (Giudice and Kao, 2004; Bulun, 2009; Table V). These molecular abnormalities are amplified when shed endometrium attaches to mesothelial cells. These changes serve not only to enhance implant survival, but also may create a local increased production of factors that promote nerve sprouting and peripheral sensitization. A reduction in progesterone action in endometrium, ‘progesterone resistance’, of women with endometriosis (Osteen et al., 2005; Table V) may play a key role in development of endometriosis and possibly with associated pain; use of progestagens may reverse these effects.
An increase of nerve fibers in the endometrium and myometrium was observed in women with endometriosis undergoing endometrial biopsy or hysterectomy compared with other women and women with endometriosis using hormonal treatments (Tokushige et al., 2006a; Al-Jefout et al., 2007; Tokushige et al., 2008; Schulke et al., 2009). This observation was studied further by two groups, who both found that the presence of nerve fibers in endometrium serves as a marker for endometriosis with high specificity and sensitivity (Al-Jefout et al., 2009, Bokor et al., 2009; Table V). The question then arises as to how this potential diagnostic tool relates to endometriosis-associated pain. In Al-Jefout's study (2009), many factors were examined for their relation to nerve fiber density without performing multivariate comparisons controlling for how these factors might interrelate. It is impossible to determine how many women had pain or how the diagnosis of endometriosis was made (biopsy or visualization), both of which would inform how nerves would serve as a diagnostic tool. Comparison of these factors considering endometriosis (as present or absent) and pain (as present or absent) was not performed, which would also have been informative. In Bokor's study (2009), a serious limitation is that few women had pain (only five of 40 women: three with endometriosis, two controls).
Together, the two studies may indicate that nerve fibers in eutopic endometrium are diagnostic of endometriosis regardless of whether the patient suffers pain. Finding nerves in the functional layer of the endometrium that is shed monthly is a provocative idea that needs replication by others. Currently, however, how endometrial nerves relate to pain specifically associated with endometriosis is unknown because nerve fibers have been found in the uterus of women with pain who do not have endometriosis (Zhang et al., 2009). How to use the information clinically is therefore uncertain. One testable possibility is that nerve growth may be hormonally modulated in eutopic endometrium and contribute to dysmenorrhea. A putative reduction in these fibers by use of progestin-dominant hormonal therapy would then present a rationale for their effectiveness in dysmenorrhea.
Retrograde menstruation is the mostly widely believed theory for how endometriosis develops. As most women have retrograde menstruation, genetic, endometrial (via changes in gene expression, hormone-induced receptor changes or other factors), inflammatory and autoimmune factors have been investigated to explain why some women have endometriosis and some do not, but no definitive risk factor has been found (Giudice and Kao, 2004; Bulun, 2009).
Women with endometriosis report having heavier menses than those without endometriosis (Treloar et al., 1998; Cramer and Missmer, 2002; Table V). Finding prostaglandins in menstrual effluent with higher levels in women with heavy menses and dysmenorrhea provides a rationale for using NSAIDs to treat dysmenorrhea (Rees et al., 1984a, b; Baird et al., 1996). These prostaglandins, if untreated, could contribute to further inflammation, providing conditions important for perpetuating sensitization. Women with genital tract anomalies obstructing the outflow of menses (Olive and Henderson, 1987; Ugur et al., 1995; Nawroth et al., 2006) frequently have endometriosis and pelvic pain (Table V). When the obstruction is removed, often lesions and pain resolve (Sanfilippo et al., 1986). Similarly, intramural and submucosal uterine fibroids, endometrial polyps, menstrual collection devices and tubal ligation can cause mechanical obstruction to menstrual flow, and thus may contribute to the pain experienced. No studies address the relationship between cervical or tubal obstruction and pelvic pain, but case reports document relief of symptoms after removal of various obstructions, suggesting an association (Morrissey et al., 2002; Spechler et al., 2003).
Uterine leiomyomas and adenomyosis are common among women of reproductive age and are associated with pain and heavy bleeding (Spies et al., 2002). Because these conditions may co-exist in women with endometriosis, attributing pain to leiomyomas, adenomyosis or endometriosis is challenging, especially because consensus non-surgical diagnostic criteria for adenomyosis using ultrasound and Magnetic Resonance Imaging (MRI) have not been developed (Vercellini et al., 2006c; Table V). Following surgical excision of endometriosis, Parker et al. (2006) report that CPP is significantly more likely to persist with junctional zone thickness >11 mm on preoperative MRI, suggesting that myometrial junctional zone abnormalities or adenomyosis contribute to CPP in women with endometriosis.
Quinn et al. hypothesize that abnormalities in innervation of eutopic uterus contribute to CPP such that small-diameter nerve fibers are present in myometrium of women with pain (but no endometriosis) and around adenomyotic nodules in those with adenomyosis, but not those without pain (Quinn and Kirk, 2002; Quinn and Armstrong, 2004). Others report increased nerve growth in myometrium in those with pain with or without endometriosis compared with healthy women (Atwal et al., 2005; Zhang et al., 2009).
The non-pregnant uterus contracts throughout the menstrual cycle (Martinez-Gaudio et al., 1973; Chalubinski et al., 1993; Bulletti et al., 2004). In healthy women, contractions propagate from fundus to cervix during menses, which presumably aids in emptying the uterus of menstrual blood (Martinez-Gaudio et al., 1973). Myometrial contractions do not occur in women on oral contraceptives compared with cycling women (Maslow and Lyons, 2004; Kido et al., 2007) suggesting a hormonal treatment effect on contractions. Since muscle contractions can be increased by prostaglandin E2 (PGE2) and prostaglandin F2α (PGF2α), and blunted by use of NSAIDs (Maslow and Lyons, 2004), prostaglandins play a role in myometrial contractility, independent of their role in neural sensitization.
Contraction patterns may differ in women with endometriosis, leading to hyperkinetic or dyskinetic contractions that impede emptying of menstrual blood or contribute to dysmenorrhea. Mean contraction pressure and frequency is higher and more frequent, and results in more retrograde endometrial debris in women with endometriosis than controls (Salamanca and Beltran, 1995; Bulletti et al., 1996; Leyendecker et al., 1996). This increase in retrograde menstrual debris may deposit more factors that could either induce nerve sprouting or contribute to neurogenic inflammation, and thus lead to continued sensitization.
Women with endometriosis usually have regular menstrual cycles of 25–37 days. In contrast, women who do not have regular menstrual cycles, like those with polycystic ovarian syndrome, usually do not report CPP or have endometriosis (Babaknia et al., 1976).
Ovulation is a complex event regulated by the hypothalamic-pituitary ovarian axis and also involves local chemokines and other factors related to inflammation (Buscher et al., 1999; Townson and Liptak, 2003; Table V). Within dominant follicles, the steroid hormone level is 100-fold higher than in plasma (Baird and Fraser, 1974). At ovulation, cytokine- and hormone-rich fluid is released into the peritoneal cavity, with levels of hormones and other factors significantly higher than in plasma (Scheenjes et al., 1990, 1991). The amount of fluid and hormones released is greatest midcycle with progesterone levels 60-fold higher and estrogen levels about 10-fold higher than in serum in the luteal phase (Donnez et al., 1982). Fluid volume and hormones released are lower in women taking contraceptive hormones, even with progestin-only contraceptives that do not suppress ovulation (Kim-Bjorklund et al., 1991). This hormone- and cytokine-rich fluid released into the peritoneal cavity with ovulation may contribute to ovulation-related pain in women with endometriosis through neural actions.
Women with endometriosis have a higher rate of luteinized unruptured follicles (LUF) than women without endometriosis (Dmowski et al., 1980; Koninckx et al., 1998). Although theorized to contribute to infertility in endometriosis, LUF could also contribute to pain because there is no rise in peritoneal levels of progesterone.
Pelvic peritoneal fluid is an ovarian exudate arising from both vascular leakage and release of fluid at ovulation (Koninckx et al., 1998). This fluid contains many macrophages and their secretion products including cytokines, interleukins and others. Endometriosis is associated with a sterile peritoneal inflammation that results in activation of macrophages that release secretory products, such as cytokines, growth and angiogenic factors (Weinberg et al., 1991; Ryan and Taylor 1997). The role of peritoneal fluid in endometriosis is described by Koninckx et al. (1998), but it is reasonable to consider that these factors in fluid potentially stimulate sprouting of nerve fibers, promote their sensitization and further increase release and activation of pro-inflammatory mediators.
Some studies suggest that progestins can normalize peritoneal fluid characteristics, which may decrease factors that enhance sensitization and pain. In vivo, dienogest decreases peritoneal fluid macrophage IL-1β production and angiogenesis, normalizes peritoneal fluid cell number and increases peritoneal fluid cells natural killer activity (Katsuki et al., 1998; Nakamura et al., 1999).
After pregnancy, pain often abates (Bulletti et al., 2004). Previously, some theorized that elevated hormones during pregnancy somehow contribute to resolving lesions, which led to use of high dose oral contraceptives to simulate a pseudopregnancy (Olive and Pritts, 2001; Table V). However, Bulletti et al. (2009) recently reported that vaginal delivery itself, rather than high hormone levels in pregnancy, may underlie symptom resolution. This observation comes from a cohort of 350 women with known stage II–IV endometriosis, dysmenorrhea and infertility. Vaginal delivery was associated with a larger internal cervical os diameter than delivery by Cesarean section or women who remained infertile. This wider cervical canal was associated with lower rates of dysmenorrhea and endometriosis recurrence than in the others, an observation that needs confirmation (Table V). Further support for the effects of vaginal delivery is that multiparity is associated with a lower risk of endometriosis (Missmer et al., 2004).
The ASRM developed a classification system that enables surgeons to describe where lesions are located on reproductive organs, extent and depth of lesions and associated adhesions involving the ovary, fallopian tube or cul de sac (Revised American Society for Reproductive Medicine classification of endometriosis, 1997). This classification system documents surgical findings, but is not useful for predicting outcomes related to infertility or pain (Schenken, 1998; Adamson and Pasta, 2009; Vercellini et al., 2006a) for three reasons. First, pain information is not noted on the form and thus comparison between pain location and lesion location is inconsistently recorded. Second, most clinical studies of pain and endometriosis summarize the ASRM classification as a measure of disease extent, but do not provide details about lesion location. Third, change in endometriosis score has been used to assess treatment effect with the assumption, but no evidence, that lesion score correlates with pain score (Shaw, 1992; Wheeler et al., 1992; Wright et al., 1995; Vercellini et al., 2009b).
Pain is a subjective and complex experience, whose understanding requires good descriptions of its individual characteristics for each patient (Dworkin et al., 2005). Despite this complexity, most endometriosis studies classify dysmenorrhea, dyspareunia and CPP as either absent, mild, moderate or severe, and describe the degree of debility experienced (Biberoglu and Behrman, 1981). This incomplete assessment strategy hampers assessment of pain outcomes (Vincent et al., 2008) and fails to consider diagnostic subtypes of CPP (Leserman et al., 2006), such as those noted with different locations of DIE (Fauconnier et al., 2002). There are efforts underway to standardize the outcome end-point in these studies with the use of visual analogue scales, but other types of scales may be needed (Dionne et al., 2005; Vincent et al., 2008).
Before pain can be attributed to endometriosis, differential diagnosis of other causes of pelvic pain should be considered. It is important that pelvic infection and diseases that are primarily myofascial, genitourinary, gastrointestinal and neurologic be investigated and excluded. Such exclusion is necessary to avoid assuming that endometriosis lesions found at surgery are the cause of the woman's pelvic pain.
An additional confounder is that women with endometriosis have a higher rate of other pain syndromes associated with peripheral and central sensitization such as irritable bowel syndrome, painful bladder syndrome, fibromyalgia and migraine headache (Gao et al., 2006; Mirkin et al., 2007; Theoharides, 2007; Tietjen et al., 2007; Vercellini et al., 2009c). In addition, women experiencing chronic pain who have endometriosis may be prone to depression, anxiety and chronic fatigue as seen with chronic fatigue syndrome (Waller and Shaw, 1995; Sinaii et al., 2002). Although the association of endometriosis with these other conditions has not been systematically studied, recognition that there are co-morbid pain conditions, and that, at different times in a woman's life, one type of pain may dominate over another, and that discomfort associated with pain attributed to endometriosis, such as dysmenorrhea, may be worsened with concomitant bladder symptoms suggests a neurologic overlap. Furthermore, in most women with endometriosis and CPP, there appears to be some concomitant myofascial dysfunction, with abdominal muscle trigger points most indicative of internal pathology (Jarrell, 2004). Muscle spasm findings are similar to those found in the ENDO rat model (Nagabukuro and Berkley, 2007b) and may signify an important facet of pain associated with endometriosis.
Women with endometriosis suffer from a wide spectrum of different types of pain, ranging from ‘simple’ severe dysmenorrhea to chronic pelvic and other co-morbid pain conditions. Current analgesic, anti-inflammatory, surgical and hormonal treatments remain unsatisfactory likely because treatments focus on treating or eliminating the growths.
Many factors can contribute interactively to individual variability in endometriosis-associated pain. Research so far encourages consideration of endometriosis as a chronic inflammatory disease, which, in turn, encourages a multi-therapeutic approach to its treatment (Table III). Furthermore, because pain of any type resides in nervous system activity, research also suggests that a major contributing factor for endometriosis-pain is not the ectopic growths themselves, but rather how sensory and autonomic nerve activity from nerves that have sprouted from nearby tissues to innervate the growths affect activity of neurons in the spinal cord and brain. Many interacting elements influence pain experienced in women with endometriosis. Some influences include menstruation, uterine and ovarian functions, estrogens, progestins, vaginal delivery, parity, peritoneal fluid, lesion type and lesion location, all of which can affect whether, when and how innervation of growths becomes sensitized, which, in turn, can affect CNS neurons and their peripherally dependent or peripherally independent sensitization (Fig. 1). A major challenge for future research is to understand how this independence develops in some women, but not others.
The most important message from this literature review, however, is that major advances in improving understanding and alleviating pain in endometriosis will likely occur if the focus changes from lesions to pain: how endometriosis fits within the knowledge of mechanisms that underlie chronic pain conditions in general. Improvements will need to be made in how endometriosis is classified in the context of pain symptoms and how pain is assessed in clinical studies and the clinic. In so doing, novel directions and strategies for treatment may become more apparent as well as how to use these treatments more effectively within a multi-therapeutic strategic approach targeted at the individual's symptom complex (Berkley and Stratton, 2009; Table III).
Overall, therefore, one research area important for improving understanding of the association between pain and endometriosis concerns the dynamics and activational physiology of peripheral nerves that supply lesions. As described here many factors likely contribute. Fundamental for advancing knowledge is recognition that engagement of the CNS by these peripheral neural processes within or near lesions and the resultant potential independence of the CNS from those peripheral processes is pre-eminent, and that research to improve understanding how endometriosis influences the spinal cord and brain of women to evoke pain is badly needed. Thus, CPP associated with endometriosis, is best considered in the context of chronic pain in general. Indeed, one growing area in the field of chronic pain is to understand commonalities across the many different forms of chronic pain (Mayer and Bushnell, 2009).
Supported by the Intramural Program, Program in Reproductive and Adult Endocrinology, Eunice Kennedy Shriver National Institutes of Health (P.S.) and NIH grant RO1 NS11892 (K.J.B.)
Alan Decherney, M.D. for his support in our writing this manuscript, Alan Hoofring of NIH Medical Arts for illustrating the figure in the manuscript and Daniel Martin, MD, Ninet Sinaii, Ph.D., MPH and Edson G. Case, JD, MA, for helpful comments during preparation of the review.