Diagnostic cut off point
References
>1 HPF
[84]
>0.5 LPF (1 mm)
>1 medium-power field (×100)
[32]
>1 LPF
[85]
>1/4 of total uterine wall thickness
[37]
2.5 mm or more
[56]
3 mm or more
[11]
Table 2.2
The classification of adenomyosis as proposed by Bird et al. [12] based on depth and extent of involvement
Depth of “invasion” | |
Grade I | Sub-basal lesions within one LPF |
Grade II | Up to mid myometrium |
Grade III | Beyond mid-myometrium |
Degree of involvement | |
Slight | 1–3 glands/LPF |
Moderate | 4–9 glands/LPF |
Marked | 10 or more glands/LPF |
In the study by Bird et al. (1972), adenomyosis (including sub-basal disease) was the sole pathology (termed ‘essential’ adenomyosis) in 92 (46 %) out of the study population of 200 cases and was the sole pathology in 75 % of cases of adenomyosis [12]. There were 47 women who had sub-basal (grade I) adenomyosis and no other pathology. Out of this subgroup, 60 % had significant menorrhagia. The incidence of menorrhagia was higher in women with sub-basal disease when compared to women with grade II and grade III lesions (n = 45) where the incidence of menorrhagia was 42 %. Thus this finding does not support definitions that exclude sub-basal lesions. Dysmenorrhoea, on the other hand, was related to the depth and the degree of involvement. The degree of involvement has rarely been the focus of research and even if reported it has seldom been included in statistical analysis. One possible explanation is that most published studies have relied on routine histology which does not regard the degree of involvement as prognostically relevant. The uncertainty linked to the appropriate cut off point lends considerable support to the suggestion made by McCausland and McCausland (1998) that histopathology should report on the actual depth of glandular presence rather than attempt a dichotomous diagnosis into normal and adenomyosis using arbitrary cut-off points [71].
Imaging Diagnosis
There are no symptoms or physical signs that are specific to adenomyosis. Classically, the uterus with adenomyosis is described as tender and symmetrically enlarged. It is interesting to note that the debate about whether adenomyosis has any characteristic symptoms is longstanding. Cullen, among other early investigators, believed that in contrast with early stage disease which is difficult to detect, fairly advanced disease could be diagnosed with great ease including by the ‘hospital assistant’. Lockyer (1918) on the other hand observed that: “it is, however, clear that in many cases, if not in most, the diagnosis is made at the operation or by the microscope” [59]. This led Lockyer (1918) to conclude that “we are therefore obliged to accept the view that an opinion expressed before operation only amounts to a probability” [59]. There is agreement in more recent literature that the specificity of preoperative diagnosis based on clinical features is poor [12], with a reported range of 2–26 % [7, 78, 85].
The introduction of transvaginal ultrasound offered an opportunity to improving the diagnostic accuracy. But earlier attempts at preoperative diagnosis using ultrasound were hampered because of the inability to reliably distinguish these lesions form fibroids [4]. The advent of transvaginal ultrasound provided a breakthrough as it was linked to improved sensitivity and specificity of >80 % (Table 2.3). The ultrasound features linked to adenomyosis include uterine enlargement in the absence of fibroids, asymmetric thickening of the anterior or posterior uterine wall, lack of contour abnormality, lack of mass effect, heterogeneous poorly circumscribed areas within the myometrium, anechoic myometrial blood-filled cysts, increased echogenecity of the endometrium, and subendometrial linear striations. Ultrasound could also detect adenomyosis as localised non-homogenous lesions within the myometrium. There is disagreement in published literature on the diagnostic value of each of these features. Meredith et al. (2009) analysed data from 14 selected published studies on the use of preoperative ultrasound and compared the findings to histological diagnosis [74]. They reported that adenomyosis was more common in women with heavy bleeding (31.9 %) compared to all hysterectomies (25.9 %). The probability of adenomyosis in a woman with heavy bleeding and positive ultrasound features was 68.1 %, compared to 65.1 % probability in a woman with positive ultrasound undergoing a hysterectomy for any symptom. But the probability of adenomyosis after a normal transvaginal ultrasound scan was 10 % in symptomatic patients compared to 8.7 % probability for women undergoing hysterectomy for any reason. The sensitivity and specificity for symptomatic women was 84.3 and 82.3 %, and for all women undergoing hysterectomy was 81.1 and 85.1 % (Table 2.4). The figures lend support to the conclusion that transvaginal ultrasound scan is an accurate test for adenomyosis, but this is necessarily weakened because of the lack of uniform histopathological or ultrasound based diagnostic criteria. This is particularly important given that most studies have used histopathology as the gold standard.
Table 2.3
The diagnostic accuracy of Transvaginal Ultrasound (TVU) and Magnetic Resonance Imaging (MRI) in various studies
Study | Sensitivity (%) | Specificity (%) |
|---|---|---|
Accuracy of TVU in diagnosis of adenomyosis | ||
Fedele et al. (1992) [28] | 87 | 99 |
Ascher et al. (1994) [2] | 53 | 67 |
Brosens et al. (1995) [15] | 86 | 50 |
Reinhold et al. (1995) [90] | 86 | 86 |
Atzori et al. (1996) [5] | 87 | 96 |
Koçak et al. (1998) [47] | 89 | 88 |
Bromley et al. (2000) [14] | 84 | 84 |
Bazot et al. (2001) [8] | 65 | 98 |
Dueholm et al. (2001) [26] | 63 | 65 |
Accuracy of MRI in the diagnosis of adenomyosis | ||
Mark et al. (1987) [65] | 61 | 100 |
Ascher et al. (1994) [2] | 88 | 66 |
Reinhold et al. (1996) [91] | 89 | 89 |
Bazot et al. (2001) [8] | 78 | 93 |
Dueholm et al. (2001) [26] | 70 | 86 |
Table 2.4
The diagnostic accuracy of preoperative ultrasound in women undergoing hysterectomy in relation to presenting symptoms [74]
Variable | All studies, n (95 % CI) | Hysterectomy in symptomatic patients, n (95 % CI) | Hysterectomy for any reason, n (95 % CI) |
|---|---|---|---|
Sensitivity | 82.5 (77.5–87.9) | 84.3 (76.3–93.2) | 81.1 (74.5–88.2) |
Specificity | 84.6 (79.8–89.8) | 82.3 (72.5–93.5) | 85.1 (79.3–91.4) |
Likelihood ratio of positive results | 4.7 (3.1–7.0) | 4.1 (2.0–8.2) | 5.1 (2.3–8.7) |
Likelihood ratio of negative results | 0.26 (0.18–0.39) | 0.25 (0.14–0.43) | 0.28 (0.17–0.45) |
Champaneria et al. (2010) published a systematic review including a meta-analysis of published articles that compared the diagnostic accuracy of transvaginal ultrasound (TVU) or MRI and that used histological diagnosis as the gold standard comparator [17]. The selection criteria for the systematic review were studies that involved premenopausal women (although studies often included both pre- and post- menopausal women) and that used the same individuals for the test and subsequently had a hysterectomy which enabled histological diagnosis. Initially, the systematic search identified 23 articles that met these selection criteria. However, 17 of identified studies were excluded because they were judged to be of poor quality, were partially duplicated with other published research or because published details were insufficient for the construction of comparison 2×2 table. This left only 3 studies that reported on the use of MRI [8, 26, 91] and 6 studies that reported on the use of TVU [8, 9, 26, 42, 91, 111]. In these studies, the pooled sensitivity and specificity of TVU was 72 % (95 % CI 65–79 %) and 81 % (95 % CI 77–85 %) respectively. TVU had a positive likelihood ratio of 3.7 (95 % CI 2.1–6.4) and a negative likelihood ratio of 0.3 (95 % CI 0.1–0.5). The pooled sensitivity and specificity for MRI were 77 % (95 % CI 67–85 %) and 89 % (95 % CI 84–92 %) respectively. MRI had a positive likelihood ratio of 6.5 (95 % CI 4.5–9.3), and a negative likelihood ratio of 0.2 (95 % CI 0.1–0.4) [17].
Despite the apparent favorable diagnostic statistics, there are many important differences between these studies. The first difficulty concerns the point discussed earlier about the cut-off point for histological diagnosis. Histological diagnosis was critical to the inclusion criteria as it was used as the reference point, but it is difficult to establish whether the cut-off points were equivalent. Bazot et al. (2001, 2002) used 2.5 mm as their cut-off point [8, 9], Dueholm et al. (2001) used a medium power field (×100) or 2 mm [26], Vercellini et al. (1998) used half a low power field (or 2.5 mm) [111], and Reinhold et al. (1996) described using one high power field [91]. The number of sections examined also varied between the studies and whilst some studies described assessment of uterine weight and morphological descriptors such as uterine wall thickness, no account is provided as to whether or how this was taken into consideration. Histological criteria recorded in a number of studies include the grade of lesions based on the depth of presence of glands and stoma within the myometrium, and lesion density [8, 9, 42]. But again, where these were mentioned as part of the methodology, they were not taken into account in the analysis. Also, those studies that made a histological distinction between focal and diffuse adenomyosis [8, 9, 26, 42] did not take this factor into account when the results were analyzed. All the studies included in the meta-analysis by Champaneria et al. (2010) included women who were scheduled for hysterectomy, but there are indicators of differences between the study populations [17]. It should be considered that the threshold for hysterectomy varies based on the population and the health care system and this may have been a factor why the incidence of adenomyosis varied between the studies ranging from 21 [26] to 37.1 % [42]. All of the 6 studies reported positive and negative predictive values which can be affected by the incidence in the population considered. It is notable that the positive predictive value of ultrasound was low (below 55 %) in all three studies included in the meta-analysis [26, 42, 111].
Also, despite the strict exclusion criteria, the two studies by Bazot et al. (2001, 2002) included 120 and 129 women respectively and described different patient profile [8, 9]. Yet both studies reported that they recruited consecutive women from the same hospital during the same time interval. This suggests significant overlap. Bazot et al. (2001) reported that 61 of the 120 patient had menorrhagia and 32 had endometrial cancer compared to the report by Bazot et al. [9] where 92 out of 129 women had menorrhagia and 13 had endometrial cancer [8, 9]. Reinhold et al. (1996) included 26 women with endometrial cancer into the study despite the fact that this can affect the appearance of the subendometrial myometrium [91]. In the study by Kepkep et al. (2007) only 8 out of 70 women had hysterectomies because of premenopausal abnormal uterine bleeding [42]. In the study by Reinhold et al. (1996) almost half the participants were postmenopausal [91]. In the study by Vercellini et al. (1998) the indication for hysterectomy was menorrhagia and/or worsening dysmenorrhoea, but the study excluded women with fibroids that distort the uterus or that were more than 12 weeks size [111]. All of the other studies included women with fibroids without stipulation of any cut off points related to the size of fibroids. A concern is that both Vercellini et al. (1998) and Reinhold et al. (1996) excluded women after investigation if the uterus could not be assessed because of fibroids although this can affect quoted sensitivity and specificity [91, 111].
With regards to image based diagnostic criteria, there was agreement on three of the features used to diagnose adenomyosis: the presence of myometrial cysts, heterogeneous myometrium and focal abnormal echotexture. In addition all studies except Dueholm et al. (2001) and Reinhold et al. (1996) included the presence of globular or asymmetrical uterus [26, 91]. Kepkep et al. (2007) and Bazot et al. (2002) but not the other studies emphasised the diagnostic value of subendometrial linear striations [9, 42]. Only Bazot et al. (2002) utilised colour Doppler [9]. Bazot et al. (2002) reported on the diagnostic value of the individual features used for ultrasound identification of adenomyosis [9]. But here again, it is interesting to note that some of these individual features had a higher sensitivity than the overall ultrasound assessment. Thus it remains unclear what relative weight was assigned by the investigators to each of the identified features. In the earlier study report by Bazot et al. (2001), the combined assessment had a higher sensitivity than the individual features [8]. The reported sensitivity, specificity, positive and negative predictive value for TVU in the study by Kepkep et al. (2007) are identical to those reported for the sonographic feature or “heterogeneous myometrium” [42]. Here again it becomes unclear what impact, if any, the other features had on the final classification.
In contrast to histopathological classification which focuses on the presence of glands and stroma within the myometrium, ultrasound diagnosis seems focussed on the appearance of the myometrium, the overall shape and size of the uterus or the presence of asymmetry. Ultrasound also emphasises the role of myometrial cysts not all of which can be histologically linked to adenomyosis. There is disagreement between the studies on whether ultrasound correctly identified the grade or degree of adenomyosis. No correlation was found between ultrasound and histopathology in the study by Bazot et al. (2001) where sonography and histopathology concurred in only 57 % of cases when assessing the depth of presence of endometrium within the myometrium and in only 23 % of cases when assessing the degree of involvement and lesion density [8]. On the other hand, Reinhold et al. (1996) reported a Kappa statistic of 0.69 indicating good agreement between TVU and histology in depicting the location of adenomyosis and a Kappa statistic of 0.81 in relation to the maximum depth of involvement [91].
Thus whilst Champaneria et al. (2010) concluded that TVU has a high level of accuracy for the diagnosis of adenomyosis, it should be borne in mind that the studies included in their review were focused on a subset of patients scheduled for hysterectomy [17]. There is lack of clarity as to the exact diagnostic criteria and the relative weight of the various features linked to the condition. In addition, the choice of histological cut off points, and the choice regarding inclusion and exclusion criteria e.g. the inclusion of women with endometrial cancer may have affected the overall assessment and ultimately the judgment in favour of TVU.
A more recent development is the advent of three dimensional ultrasound and its use in relation to adenomyosis. Little research has been published so far. Naftalin et al. (2012) reported on the use of 3D-TVU in 985 consecutive women who attended a general gynaecology clinic in a large teaching hospital in the UK [81]. They reported the prevalence of adenomyosis in the whole group as 20.9 % (95 % CI: 18.5–23.6 %). It was possible to compare histological findings with 3D-TVU in the subgroup of women (n = 45) who subsequently underwent a hysterectomy. After excluding women with cancer (n = 14) and with large fibroids (n = 4), the investigators reported a fair level of agreement between 3D-TVU and histological diagnosis of adenomyosis [κ = 0.62 (p = 0.001), 95 % CI (0.324, 0.912)]. They reported a positive correlation between age and the finding of adenomyosis, but the incidence is not provided divided by age groups or by clinical presentation. Thus although the study population may provide a range of diverse presentations, it will necessarily be affected by specialisation within the clinic and referral criteria therefore caution should be exercised when extrapolating the figures to different populations. Luciano et al. (2013) prospectively evaluated the accuracy of 3D-TVU in 54 symptomatic premenopausal women undergoing hysterectomy for benign conditions [61]. Of these, there were 32 patients who had no previous treatment, 26 of whom had adenomyosis. Features linked to adenomyosis were: (1) Maximum Junctional Zone thickness (JZmax) ≥8 mm, (2) myometrial asymmetry and (3) hypo-echoic myometrial striations. They reported that when at least 2 of these features were present, 3D-TVU was 90 % accurate (sensitivity = 92 %; specificity = 83 %; PPV = 99 %; and NPV = 71 %). Interestingly the accuracy reduced to 50 % in the subgroup who had undergone endometrial ablation (n = 12), and was 60 % in the group receiving medical treatment (n = 10).
Thus – despite much promise – studies assessing the role of ultrasound and MRI in diagnosing adenomyosis all suffer methodological weaknesses some of which are due to the constraints inherent in the study population. The need for histology as the gold standard means that only a particular cohort could be assessed. Still, a major difficulty has been in the use of retrospective cohorts which suffer from lack of standardisation and incomplete assessment. The indications for hysterectomy are becoming increasingly narrow which add to the need for reliable non-invasive diagnostics.
Biomarkers in Adenomyosis
Despite the breakthrough achieved with the use of MRI and transvaginal ultrasound, reliable diagnosis of adenomyosis remains difficult and expensive. Thus, the identification of a non-invasive reliable marker for the disease will have significant clinical value. Such a marker may also help monitor disease progression or response to treatment.
CA125 is perhaps one of the earliest biomarkers to be studies in relation to endometriosis and adenomyosis. CA125 is produced by most non-mucinous epithelial ovarian tumours. More research has been directed to assessing its use in endometriosis than to adenomyosis, but meta-analysis of published results concluded that it is of limited utility [77]. More recent research has again demonstrated the limited utility of CA125 in endometriosis without endometrioma, but that accuracy could be improved by using a lower combined cut-off values for CA-125 at 20 and 30 U/mL [46]. Concomitant use of CA125, CA19-9 and IL-6 did not add significantly to the value of CA125 alone [99].
In relation to adenomyosis, Takahasi et al. (1985) examined 11 patients with fibroids, 7 with adenomyosis and 1 with adenomyosis and fibroids and reported that the mean CA125 level (± SD) was 18.3 (±6.1) U/ml in patients with fibroids and 93.3 (±49.4) U/ml in those with adenomyosis [100]. The difference was statistically significant. Seven out of the 8 women with adenomyosis but none of those with fibroids had serum CA125 > 35 U/ml. Following surgery, CA125 level in patients with adenomyosis gradually decreased and returned to normal 1 month postoperatively. But the diagnostic value was disputed by others. Halila et al. (1987) measured serum CA125 in 22 women undergoing a hysterectomy for adenomyosis or fibroids but reported normal CA125 levels (<35 U/ml) in 20 patients including in all those with histologically proven adenomyosis [35]. One complicating factor is the observation that serum levels of CA125 varies with the menstrual cycle. Masahashi et al. (1988) reported a transient rise during menstruation. They also reported that serum levels are higher in patients with adenomyosis and with advanced endometriosis compared to normal controls [67]. Takahashi et al. (1988) reported that after control for cycle phase, CA125 was elevated in patients with adenomyosis (as well as in endometriosis) [101]. Interestingly, Bischof et al. (1992) reported elevated CA125 levels in women with fibroids. They attributed this to increased peritoneal distension secondary to uterine enlargement by the fibroid [13].
Agic et al. (2008) measured chemokine (C-C motif) receptor 1 mRNA (CCR1 mRNA) in peripheral blood leukocytes together with monocyte chemotactic protein-1 (MCP-1) and CA125 protein in serum of women with endometriosis and adenomyosis. The ratio of CCR1/HPRT mRNA (Hypoxanthine-guanine phospho-ribosyl-transferase) in peripheral blood of patients with endometriosis was significantly elevated compared to women without endometriosis. No significant difference in CCR1/HPRT mRNA levels was found between women with adenomyosis and the control group. Serum levels of MCP-1 and CA125 were significantly higher in patients with endometriosis. The combined test using the three markers was considered positive if at least one of the markers was above the set threshold. When used to detect endometriosis, this combined test showed sensitivity, specificity, NPV, PPV of 92.2 %, 81.6 %, 83.3 % and 92.3 % respectively. The combined test predicted the presence or absence of adenomyosis to a lesser extent: sensitivity, specificity, NPV, and PPV were 72.7 %, 81.6 %, 93.0 %, 47.1 % respectively [1].
Another approach is the use of proteomic analysis of serum samples. Long et al. (2013) compared serum samples from women with adenomyosis, endometriosis and controls using MALDI-TOF-MS proteomic analysis. They identified 13 protein peaks that were abnormally expressed in endometriosis and 12 in adenomyosis compared with control groups. Five-peak masses were significantly down regulated both in the women with endometriosis and adenomyosis. Two protein peaks with m/z of 2.748 and 5.759 kDa were reported to be of high value in the diagnosis of adenomyosis [60]. However, this approach is fraught with difficulty. Previous studies using this technique in endometriosis have led to the identification of different putative protein markers. Jing et al. (2009) identified two marker proteins with m/z of 5.83 and 8.865 kDa [40]. Kyama et al. (2011) reported that endometriosis was diagnosed with high sensitivity (89.5 %) and specificity (90 %) with use of five down-regulated mass peaks (1.949, 5.183, 8.650, 8.659, and 13.910 kDa), and minimal-mild endometriosis was diagnosed with four mass peaks (two up-regulated: 35.956 and 90.675 kDa and two down-regulated: 1.924 and 2.504 kDa) with maximal sensitivity (100 %) and specificity (100 %). The 90.675 and 35.956-kDa mass peaks were identified as T-plastin and annexin V [52]. Ding et al. (2010) detected 3 mitochondrial protein peaks as potential biomarkers for endometriosis with m/z of 15.334, 15.128 and 16.069 kDa [24]. The differences may be due to different experimental conditions, different protein chips or technologies used, or to patient related factors.
Xiaoyu et al. (2013) used iTRAQ (isobaric tags for relative and absolute quantitation) technology to compare serum samples from women with and without adenomyosis. They reported that 21 proteins were significantly up-regulated and 4 proteins were significantly down regulated in women with adenomyosis (Table 2.5) [119]. They thus raised the possibility of using the identified proteins as biomarkers for adenomyosis.
Table 2.5
Proteins differentially expressed when comparing serum samples from women with adenomyosis and controls using iTRAQ analysis [119]
Up–regulated proteins | |
Fibrinogen α | Fibrinogen β |
Fibrinogen γ | CD44 |
Fibronectin 1 | Complement C1r |
Apolipoprotein B-100 | Complement factor B |
Hemoglobin subunit δ | Complement C1s |
Complement C3 | Complement C5 |
Antithrombin-III | Vitamin K-dependent protein S |
Ceruloplasmin | Serum amyloid P-component |
Leucine-rich α-2-glycoprotein | α-1-antichymotrypsin |
Inter-α-trypsin inhibitor heavy chain H4 isoform 1 | Vitamin D-binding protein |
Apolipoprotein C-II | |
Down–regulated proteins | |
Gelsolin isoforms-a | Apolipoprotein A-IV |
Transthyretin | Keratin, type I cytoskeletal 9 |
Dechaud et al. (2014) performed gene expression array in adenomyosis and reported that the most up regulated genes in the endometrium in adenomyosis were SH2D3A, KLHL31 and ADAMTS16 whilst the most down regulated genes were FOXP2, F2RL2 and DGKB and raised the possibility of these being useful as markers of adenomyosis [23].
Clinical Manifestations of Adenomyosis
As mentioned above, the preoperative diagnosis of adenomyosis is poor. In one study, the diagnosis was suspected preoperatively in only 10 % of cases and recognized at surgery in 35 % of patients [85]. It is perhaps well recognized that there are no symptom or symptoms that are individually or collectively pathognomonic of uterine adenomyosis. Traditionally adenomyosis has been linked to a variety of common gynaecological presentations, most prominently abnormal bleeding, dysmenorrohea and although it is more common in parous women, it has been linked to infertility. It is also recognised that many cases are identified in asymptomatic women. This will be explored further, but it is important to point out that a variety of other gynaecological conditions such as endometriosis and fibroids have also been linked to these presentations as well as being diagnosed in asymptomatic women. Both endometriosis and fibroids are commonly present in association with adenomyosis. The significance of the finding of adenomyosis needs to be considered against the knowledge that the threshold at which women seek medical care for any of these presentations varies and at the same time, the clinical threshold for defining normality is not always clear or agreed.
Symptoms Linked to Adenomyosis (Box)
Abnormal Uterine Bleeding
Heavy menstrual bleeding is one of the more common indications for hysterectomy, and as adenomyosis has been reported in a sizable percentage of surgically removed uteri, it is not surprising that heavy menstrual bleeding has come to be linked to adenomyosis. In the study by Bird et al. (1972) 200 hysterectomy specimens were assessed for the presence of adenomyosis. Lesions were classified into three grades: Grade (I), sub-basal adenomyosis where the lesions were found within one low power field below the basal endometrium, but no further; Grade (II), where adenomyosis was found up to the mid-myomstrium; and Grade (III) where adenomyosis extended beyond the mid-myometrium [12]. Adenomyosis was identified histologically in 31 % of the 200 specimens examined using routine histopathology, but when additional sections were taken, 38.5 % were identified as having adenomyosis and the figure rose to 61.5 % when Grade I (sub-basal) adenomyosis was included. Adenomyosis was the only uterine lesion in 16.5 % of cases and was the major pathology found in 32.5 % of cases. In 46 % of all cases (92 out of the 200 women included in the study) adenomyosis was either present alone or together with other non-significant pathology, this included 47 Grade I, 33 Grade II, and 12 Grade III cases. Thus all the 47 cases of sub-basal adenomyosis belonged to the group where adenomyosis was the sole significant pathology. Ninety (83.5 %) of the women identified with adenomyosis (n = 123) had associated pathology. These included fibroids (n = 68), endometrial hyperplasia (n = 9), endometriosis (n = 8), or polyps (n = 5). The presence of pathology associated with adenomyosis is well recognised in literature. Of the 92 women who had adenomyosis alone or with no other significant pathology in the report by Bird et al. (1972), 51.2 % had menorrhagia, 10.9 % had metrorrhagia, 28.3 % had dysmenorrhoea, 2.2 % had postmenopausal bleeding and 23.9 % were asymptomatic [12]. Only 18.7 % had both menorrhagia and dysmenorrohea. Of the 47 patients who had adenomyosis sub-basalis, 60 % had significant menorrhagia compared to 19 (42 %) of the 45 women who had grade II or III adenomyosis. Thus the difference between the two is not statistically significant. Two of the 47 patients with Grade I disease had dysmenorrhea, compared to 14 of 33 with Grade II, and 10 of the 12 women with grade III. In terms of the degree of involvement, dysmenorrhea was present in 13.3, 26.7, and 58.8 % of women with slight, moderate, or marked disease. The difference was statistically significant. Although Bird et al. (1972) did not provide information about how symptoms or symptom severity were assessed or a definition of what constituted metrorrhagia, they concluded that adenomyosis “may cause hypermenorrohea and increasingly severe, acquired dysmenorrhea” [12].
Box Symptoms Linked to Adenomyosis
Menorrhagia | Increased |
Dysmenorrhoea | Increased |
Chronic pelvic pain | Increased |
Dyspareunia | Limited data |
Infertility | Increased |
Spontaneous abortion | Increased |
Owolabi and Strickler (1977) used one LPF as a cut-off point and used two random tissue blocks in routine histopathology to diagnose adenomyosis and identified adenomyosis in 161 out of 1619 (10 %) consecutive hysterectomies [85]. In 97 (60.2 %) cases, there was coexistent pathology, mostly fibroids, endometrial hyperplasia and carcinoma, and endometriosis. They reported that 65 % of the group who had adenomyosis as the sole pathology (n = 64) had abnormal bleeding and that there were also symptoms of dysmenorrhea, non-menstrual pelvic pain and/or dyspareunia. It is not possible to understand these figures further as the exact number of patients affected is not provided and the article reports individual symptoms rather than patients affected. None of this group was asymptomatic, but two of those with abnormal bleeding were postmenopausal with atrophic endometrium and their symptoms are thus unlikely to be related to adenomyosis. In addition, there were five asymptomatic women who had associated pathology (three had CIN and two had adnexal masses) and thus no symptoms attributable to the presence of adenomyosis. It remains speculative if non-menstrual pain or dyspareunia that was present in 12 and 6 % of the adenomyosis only group can in fact be attributable to adenomyosis. Although the study concludes that the presence of adenomyosis is always associated with symptoms it should be considered that the group as a whole were symptomatic, hence the hysterectomy, but also that the study does not provide a comparison with patients who underwent hysterectomy but did not have adenomyosis. Furthermore, it is possible that a significant number of cases with adenomyosis were missed because of the sampling protocol that was followed.
Levgur et al. (2000) assessed 111 uteri all of which were below 280 g for the presence of adenomyosis [56]. When present, the lesions were classed as superficial if they were at a depth of less than 40 % of the uterine wall, intermediate if they were found at a depth between 40 and 80 % of uterine wall and were classed as deep if they were present at more than 80 % of uterine thickness. The authors reported an association between the number of foci and the depth of endometrial presence within the myometrium. The median number of foci was higher in women with dysmenorrhoea compared to those without dysmenorrhoea, but there was no difference in the number of foci in women with or without menorrhagia. In this study, superficial-depth-adenomyosis was not associated with menorrhagia or with dysmenorrhea. However, Levgur et al. (2000) excluded from the definition of adenomyosis lesions that were less than 2.5 mm below the endometrium [56]. Also excluded were 132 women who had a uterus >280 g in weight. The given reason for the exclusion was the difficulty obtaining full thickness myometrial biopsies (this group included 6 women with adenomyosis). In the 111 women included in the study, 17 had adenomyosis alone, 19 had adenomyosis and fibroids and 39 had fibroids but no adenomyosis. No information is provided on other associated pathology, or on the indications for hysterectomy. The authors state that menorrhoagia and dysmenorrhoea were associated with ‘degree of myometrial depth’ and that menorrhagia occurred in 36.8 % of women with deep foci and 13.3 % with intermediate foci. The corresponding figures for dysmenorrohea were 77.8 % and 12.5 % respectively. However, it is difficult to assess the significance of the findings as the figures were only provided as percentages and it is not stated whether the denominator included all women with adenomyosis or whether that was restricted to the subgroup without fibroids. Also, while the study objective was to correlate symptoms of uterine adenomyosis with histopathologic findings, the number of women with menorrhagia, dysmenorrohea or both and the indications for hysterectomy are not provided. The age ranges suggest that a large percentage were postmenopausal. Other methodological problems include the relatively small number of slides examined per patient, and that lesions were reported as number per sections examined rather than as lesion density.
Sammour et al. (2002) examined 94 uteri from women who underwent a hysterectomy and who were diagnosed with adenomyosis. Twenty five of these women also had fibroids [94]. The indications for hysterectomy are not provided, but the mean ages suggest that a good proportion may have been postmenopausal. The specimens were classified into four groups each corresponding to 25 % of myometrial thickness. Foci less than 2 mm below the endometrium were not included in the definition. The four groups were compared in relation to the symptoms of menorrhagia, dysmenorrhea, dyspareunia or pelvic pain, put no difference was found between the groups. The ‘spread’ of adenomyosis was assessed by examining the number of foci per slide and the number of slides varied according to the presence or absence of gross disease. The symptoms were not defined beyond the title, thus the distinction between pelvic pain and other pain symptoms is not clear. Comparisons were made based on the main complaints, yet more than one complaint was recorded per patient. The main finding of this study was a lack of correlation between symptoms and the depth of adenomyosis and that there was a significant correlation between pelvic pain or dysmenorrhea but not between menorrhagia or dyspareunia and the ‘spread’ of adenomyosis. These findings should be interpreted with caution because of lack of standardization in defining disease ‘spread’ and because the indications for the surgery is not provided. In addition, only three tissue blocks were examined per specimen leaving the possibility of under diagnosis of adenomyosis.
Ozkan et al. (2011) reviewed the records of 1680 patients who underwent a hysterectomy [86]. Amongst this group, 98 patients were identified with adenomyosis and 106 had fibroids. Most (61 %) of the group with adenomyosis and 48 % of the group with fibroids were >50 years old. The diagnostic cut-off point and the number of tissue blocks assessed is not stated, but the overall incidence of adenomyosis in this group (12 %) was lower than reported in most other recent series. No indication is provided about associated pathology or about the number of patients with concomitant fibroids and adenomyosis. Ozkan et al. (2011) made a distinction between the frequency of dilatation and curettage – which was not statistically significantly different between the two groups – and endometrial sampling and the incidence of adenomyosis [86]. There was a higher incidence of endometrial sampling in the adenomyosis group. Therefore, Ozkan et al. (2011) argued that intrauterine sampling may trigger adenomyosis through deterioration of the endomyometrial junction [86]. But whist it is possible to speculate that deep endometrial sampling through overzealous curettage may disrupt the endomyometrial junction, modern alternative endometrial sampling techniques are designed to obtain more superficial samples of the functionalis endometrium and are unlikely to result in direct injury to deeper tissue. The more frequent resort to endometrial sampling in this group may reflect clinical practice in response to clinical presentation. In support of this is the observation that there was a statistically significant difference between the number of women undergoing hysterectomy for endometrial hyperplasia in the adenomyosis group (n = 32) and in the group with fibroids (n = 20), and a statistically significantly higher number of postmenopausal women in the adenomyosis group (n = 48) compared to the group with fibroids (n = 36). Interestingly, more than half of the patients in both groups were diagnosed with ‘tubal inflammation’, between 35 and 49 % had ovarian cysts and between 91 and 93 % had ‘chronic cervicitis’ as coexisting pathology. But there was no mention of other pathologies known to be associated with adenomyosis such as polyps or endometriosis. In their binary logistic regression Ozkan et al. (2011) identified age, menometrorrhagia and endometrial sampling as important covariant associated with adenomyosis [86]. However, the incidence of menometrorrhagia in the adenomyosis group (35 %) was lower than the incidence in the group with fibroids (43 %) and examination of menstrual bleeding was limited to a classification into 4 groups: regular, oligomenorrhea, menometrorrhagia and menopause which may be a reflection of clinical practice where menstrual bleeding patterns and/or quantity are poorly explored.
In a retrospective case control study from the United States, Taran et al. (2010) compared women undergoing hysterectomy with adenomyosis or with fibroids as the sole pathology [102]. They identified 76 cases with adenomyosis which were matched 2:1 by surgeon and by year of surgery to 152 women with fibroids only. The rationale for matching by surgeon is stated as the elimination of confounders of referral patterns and the elimination of bias based on the effect of concomitant procedures on practice style. However, no indication is provided as to what these confounders might be, or of how matching was undertaken within the practice of each surgeon beyond the given time frame (±1 year). Of the patients identified as having had a hysterectomy during the study period (n = 1871), 582 had fibroids, 133 had adenomyosis and 53 had both. This gives a relatively low overall incidence of adenomyosis of 186 (10 %), but the diagnostic criteria used for adenomyosis are not provided. The exact ethnic distribution is not provided, but it is stated that 95.1 % of both study populations were Caucasian. The indications for hysterectomy in 92.1 % of the adenomyosis group and in 94 % of the hysterectomy group were the presence of adenomyosis or leiomyomas or the presence of one or more disease-specific symptoms. The remaining hysterectomies were performed for indications of uterine prolapse, grade II cervical intraepithelial neoplasia, endometriosis and permanent sterilization. Taran et al. (2010) identified differences in the age distribution, the group with adenomyosis being relatively younger (41 ± 6.4 years) compared to the group with fibroids (44.4 ± 4.8 years) [102]. There were no differences between the groups in the number of children, miscarriages or abortions the women had. The duration of menstrual bleeding was also similar in both groups (7.9 ± 3.6 days in the adenomyosis group and 7.9 ± 4.2 days in the group with fibroids). There was a higher incidence of depression (55.3 % vs. 26.3 %) and of the use of antidepressant in the adenomyosis group (35.5 %) compared to the group with fibroids (19.1 %). The authors put forward the suggestion of a possible aetiological link to antidepressants through an effect on raised prolactin secretion secondary to their use. There was a higher incidence of dysmenorrhea (60.5 % vs. 39.7 %), dyspareunia (17.1 % vs. 6 %) and of the use of NSAID (67.1 % vs. 42.1 %) in the adenomyosis group compared to the group with fibroids. There was also a higher proportion of women with abnormal cervical smears (30.3 % vs. 16.5 %) and of procedures for cervical dysplasia (9.2 % vs. 2 %) in the group with adenomyosis. However, the symptom complex of the two groups is necessarily affected by the indication for hysterectomy. Adenomyosis per se is not, and rarely are fibroids, an indication for hysterectomy in the absence of associated symptoms. As such, much of the quoted outcomes including abnormal smears, pain symptoms, abnormal bleeding, surgical intervention for cervical dysplasia, and endometriosis were not independent of the reason why surgery was performed. It is also possible that the presence of chronic pain was associated with the need for antidepressants. As mentioned above, the reliability of the outcome data of this and other retrospective studies will necessarily be affected by the thoroughness by which clinical detail was collected. This is not restricted to random errors, but there can be systematic points emanating from the way diseases are viewed. Whilst documentation in prospective research can be standardized between comparison arms, information available for retrospective research relies on available documentation which may vary from the most thorough to what individual clinicians may regard as sufficient. Thus the absence of documentation of any particular symptom can be open to various interpretations. In addition, the severity of documented symptoms and their clinical impact can vary considerably for a variety of reasons. It is also possible that the threshold for surgery may be lower in the presence of anatomical lesions such as fibroids. Taran et al. (2010) identified a history of infertility to be significantly linked to adenomyosis (14.1 % vs. 4.6 %) mainly because of associated endometriosis [102]. However, endometriosis was one of the quoted indications for surgery. Still, there was no difference between the adenomyosis and the fibroid groups in gravidity (2.7 ± 2.2 vs. 2.4 ± 1.8), parity (1.9 ± 1.4 vs. 1.9 ± 1.3), the number of spontaneous miscarriages (0.7 ± 1.4 vs. 0.4 ± 0.9) or of therapeutic abortions (0.1 ± 0.4 vs. 0.02 ± 0.2). It is notable that Taran et al. (2010) restricted their multivariable regression analysis to patients with symptoms of abnormal bleeding and/or pain which they believed to be ‘disease-specific symptoms’ [102]. This assumption limits the utility of this study towards addressing the basic question of whether adenomyosis is in fact relevant to these symptoms or whether it is incidental.
A different view-point was presented by Weiss et al. (2009), who reported on the findings of a study involving women who underwent a hysterectomy whilst under follow-up as part of a trial primarily concerned with the health of women during their middle years [118]. There were 3302 eligible women identified from seven centers in the US, but 200 women never completed a follow-up. At the time of recruitment women had to be aged between 42 and 52 years and to have an intact uterus. After 9 years of follow-up, 239 women underwent a hysterectomy (8 %). It was possible to obtain consent and the medical records of 137 women for the purpose of the report by Weiss et al. [118]. These were divided into two groups; one group comprised women reported as having adenomyosis on histological examination (n = 66), the other group comprised all other patients (n = 71). Case notes were obtained retrospectively and examined to compare the characteristics of both groups. The diagnosis of adenomyosis was obtained from the clinical records based on local hospital practice, but the criteria are not defined. It is notable the while adenomyosis was present in 48 % of all samples, only one patient had adenomyosis with no associated pathology. Women with adenomyosis were more likely to have been pregnant (95 %) compared to those with no adenomyosis (85 %) and the difference was statistically significant. The two groups were not statistically significantly different in factors of ethnicity, educational attainment, income category, smoking, number of pregnancies, BMI, age at hysterectomy or uterine weight. There were no differences between the two groups with regards to their symptoms at the time of hysterectomy. The most common presentations in the adenomyosis group were problems with vaginal bleeding (n = 35), fibroids (n = 34), chronic pelvic pain (n = 15), prolapse (n = 6), stress urinary incontinence (n = 5), acute pelvic pain (n = 3). The most common presentations for the group with no adenomyosis were fibroids (n = 46), problems with vaginal bleeding (n = 43), chronic pelvic pain (n = 19), prolapse (n = 7), stress urinary incontinence (n = 6) and acute pelvic pain (n = 3). As there were no statistically significant differences in the presenting diagnosis for women with or without adenomyosis, Weiss et al. (2009) argued that despite a woman’s presenting symptom or indication for hysterectomy, she is equally likely to have or not to have adenomyosis [118]. Weiss et al. (2009) identified three ‘associations’ with adenomyosis: fibroids, endometriosis and abnormal bleeding [118]. These were present in 51 (37 %), 4 (3 %) and 35 (27 %) of cases with adenomyosis, and in 59 (43 %), 7 (5 %), and 43 (33 %) of the group that did not have adenomyosis. The authors therefore argued that there was no association between the presence of abnormal bleeding or endometriosis and the presence or absence of adenomyosis. The authors also undertook a multivariate logistic regression analysis with fibroids, endometriosis, abnormal bleeding or chronic pain as independent variables to assess whether these conditions were associated with adenomyosis independent of other factors and found no association. Yet again, this study shares many of the weaknesses of the other retrospective studies published to date. There is no indication about how adenomyosis was defined and the symptoms prior to hysterectomy were only superficially described. All uterine bleeding is included under the heading of abnormal bleeding thus overlooking basic distinctions such as that between pre- and post- the menopause. In addition, fundamental problems become apparent when assessing the study design against the hypotheses being tested. Weiss et al. (2009) wrote that their study tested four hypotheses: (1) adenomyosis is associated with the presence of fibroids; (2) adenomyosis is more common in the presence of endometriosis; (3) adenomyosis is associated with abnormal uterine bleeding; (4) symptoms of chronic pain are more likely in uteri with fibroids if adenomyosis is present [118]. They concluded the data generated is their study did not provide evidence in support of these hypotheses. Testing the association with fibroids requires the assessment of uteri identified with adenomyosis for the presence of fibroids compared to a group without adenomyosis. The difficulty here is that fibroids were present as a reason for hysterectomy in the majority (n = 80 or 58 %) of the study population, yet it is included as an outcome measure. In relation to the second hypothesis, the research design does not inform what associated pathology exists in women with endometriosis. In addition endometriosis is not a disease of the fifth or sixth decades. Neither can this study design inform the debate about the symptoms that may be linked to adenomyosis or to uteri with both fibroids and adenomyosis. Thus a main flow in the study is the inclusion of entry criteria (fibroids, bleeding, and pain) as outcome variables in the analysis.
Vercellini et al. (1995) compared the incidence of adenomyosis in 1334 hysterectomy specimens in relation to the indication for hysterectomy. Adenomyosis was identified in 332 (24.9 %) of all cases [112]. The incidence of adenomyosis was 23.3 % in women with fibroids and menorrhagia compared to 25.7 % in women with prolapse, 21.4 % in women with ovarian cysts, 19 % in women with cervical cancer, 28.2 % in women with endometrial cancer, 28.1 % in women with ovarian cancer and in 24.7 % of women with other miscellaneous indications. The difference between the groups was not statistically significant. These findings, if confirmed, suggest a weaker link between adenomyosis and menstrual symptoms. The study by Vercellini et al. (1995) relied on routine histological assessment of removed samples, and used a cut-off point of half a LPF for adenomyosis (estimated to be about 2.5 mm) [112]. But again it has a number of significant weaknesses. For example, the study included some cases with malignancy which may undergo more rigorous sampling; in addition the analysis included fibroids and menorrhagia within the same analysis group without a clear rationale. The retrospective design did not allow adequate assessment of the menstrual history, or an assessment of dysmenorrhea or pelvic pain which are important outcome measures. No definition is provided of what is grouped under the heading ‘menorrhagia’, and no indication is given of the menstrual history of patients who underwent hysterectomy for other reasons. The study also suffers from incomplete ascertainment of data. For example, information about spontaneous or induced abortion is provided on 134 (40 %) and 105 (32 %) women respectively in the adenomyosis group. The corresponding figures for the group without adenomyosis were 343 (34 %), and 262 (26 %). Some of the two patient groups may have been misclassified in relation to the presence or absence of adenomyosis and the indication for surgery. It is clearly possible that adenomyosis may account for menstrual symptoms in some but not all those affected, or that some women with menstrual symptoms respond to conservative or medical treatment but undergo hysterectomies for other indications later in life.
In a subsequent study, the same group published a report on a group of women (n = 707) who underwent a hysterectomy and who had clinical information collected in advance of the operation [89]. The indications for hysterectomy were fibroids and/or menorrhagia (n = 140, 19.8 %), prolapse (n = 100, 14.1 %), ovarian cyst (n = 81, 11.5 %) or cancer (n = 14, 2 %). About a fifth of the cohort (n = 150, 21.2 %) were identified with adenomyosis using the same cut-off point as per their previous study; half a LPF or about 2.5 mm below the endometrial-myometrial junction. But no indication is provided of the incidence or the type of associated pathology in the group with adenomyosis or of the findings in the control group. Parazzini et al. (1997) reported that women who smoked were at lower risk of adenomyosis, and that the risk seemed inversely related to the number of cigarettes smoked [89]. But the age-adjusted trend in risk was of borderline statistical significance (χ 2 trend 3.57, p = 0.06). Adenomyosis was higher in parous women and in relation to number of children compared to nulliparous women (χ 2 trend 20.71, p < 0.01) and in those who had spontaneous abortions (odds ratio = 1.7; 95 % CI 1.1–2.6). There was no difference in relation to the use of oral contraception, IUCD or a history of induced abortion. Parazzini et al. (1997) stated that the risk of adenomyosis tended to be lower in more educated women but that the finding was not statistically significant [89]. The study found no difference between the two groups in the incidence of dysmenorrhea, intermenstrual pelvic pain or dyspareunia.
One of the main difficulties with the study is the challenge of controlling for confounders. There is a complex interaction between socioeconomic and demographic factors including factors such as age and parity and symptoms in decisions for hysterectomy. Much of this is now well documented. Indeed the authors attempted to control for these through the use of age and multivariate adjusted models. One analysis included controlling for age and intensity of flow in a comparison involving the menopausal status. This concluded that there was no relation between the menopausal status and the incidence of adenomyosis, although a significantly higher proportion of the group with adenomyosis were postmenopausal (48 % vs. 33.5 %, p = 0.0016). There was no difference in the incidence of heavy flow based on the presence (39.7 %) or the absence (35.4 %) of adenomyosis when the two groups were compared, but the difference was statistically significant in the age adjusted (odds ratio 1.7; 95 % CI 1.1–2.6) but not in the multivariate (odds ratio 1.4; 95 % CI 0.9–2.2) model. Indeed it is arguable that age is relevant to a number of other factors included in the analysis such as a history of dilatation and curettage which used to be a very common procedure in the past but that has now been largely abandoned. Induced abortion is far more common now compared to former years. Because of the more focused effort at recoding the menstrual history, Parazzini et al. (1997) were able to perform a more detailed analysis than was possible in older literature [89]. Despite this, it is difficult to see how the information included in the analysis could be a reflection of patients’ presentation. The main categories included in the study are described as: (1) categories based on the ‘lifelong menstrual pattern’. Menstrual history is used to categorize women into three categories based on the length of the menstrual cycle (<25, 26–30, and >31 days); (2) categories based on duration of bleeding. Here, ‘flow days per month’ was used to categorize women into two groups depending on whether their loss lasted 5 or fewer days, or >5 days; (3) categories based on amount of loss. Here, ‘intensity of flow’ was used to categorize women into two categories as being either regular or heavy. Whilst recognizing the difficulties inherent in providing an accurate description of menstrual cycles, information collected retrospectively but prior to hysterectomy could hardly provide an accurate account of lifelong menstrual patterns. Menstrual patterns are known to change overtime including in women who have no menstrual complaints. In their conclusion, Parazzini et al. (1997) stated that no relationship was found in their study between adenomyosis and several menstrual characteristics including polymenorrhea and pain and that the relationship with heavy cycles disappeared in the analysis after adjustment for potential covariate [89]. They add that the presence of endometriosis was not associated with adenomyosis. But as explored above, the design of this study is not suited to addressing the question of whether there is a relationship between adenomyosis and the symptoms described.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
