Keywords deep infiltrating endometriosis - transvaginal sonography - hormonal therapy - bowel
endometriosis - long-term observation
Introduction
The natural history of deep endometriosis in reproductive-aged women is unclear [1 ] since the pathophysiology of onset and growth are unclear [2 ]
[3 ]. Deep endometriosis lesions are less frequent and smaller in adolescence. Therefore,
the lesions must have grown after initiation, at least for a certain period of time.
However, growth itself has not been documented, and growth might be self-limiting,
as suggested by the volumes of deep endometriosis lesions at different ages [4 ].
Although imaging of DE by magnetic resonance imaging (MRI) or transvaginal ultrasound
(TVS) has been established as a reliable and accurate diagnostic method to measure
deep endometriosis [5 ]
[6 ], sequential measurements over longer periods of time in women not undergoing surgery
have not been performed, except to evaluate the effect of medical therapy.
Deep endometriosis is treated by medical therapy or surgical excision [7 ]. Medical therapy is often the first line of therapy to avoid often difficult and
complication-prone surgery. If pain symptoms are sufficiently reduced, this therapy
can be continued for longer. It was estimated that some two-thirds of patients can
be managed by medical therapy [8 ]. It is equally well documented that during medical therapy, lesions regress. Fedele
et al. observed no relevant progression of colorectal DE lesions in most asymptomatic
and untreated women over six years using transrectal sonography [9 ]. Barra and colleagues observed a significant reduction in the volume of the nodule
during the administration of dienogest [10 ]. Netter et al. demonstrated that the length of amenorrhea, induced through hormonal
intake or pregnancy, correlated with regression of the size of rectal DE [1 ].
However, endometriosis lesions are biochemically heterogeneous with a variable degree
of aromatase activity and progesterone resistance. This might explain why some 10%
of women do not respond and 20% respond poorly to medical therapy [11 ]. It is unclear whether a decrease in the nodule volume and symptom relief correlate.
Abrao et al. found that pain symptoms correlate with the sonographic dimensions of
DE [6 ]. Since some occasional nodules were observed to progress during medical therapy,
explained by the variable response and the biochemical heterogeneity, follow-up with
US was recommended [6 ].
There are no data on the growth or microscopic changes of DE managed without treatment
[10 ]
[12 ].
Since data of longitudinal follow-up range from 6 months to 10 years, we decided to
review our patients managed conservatively without surgery with a follow-up of up
to 18 years.
Patients and methods
Study design
The study was conducted as a monocentric case series of all women with deep endometriosis
of the rectum not undergoing surgery and managed conservatively. Informed consent
about the anonymous use and publication of the data was obtained.
The ethical committee approved the study (approval number S2022–16) on the August
10, 2022.
Study population and data collection:
The inclusion criteria were women with deep infiltrating endometriosis of the rectum
followed up clinically and by US by JK at the tertiary center between July 2002 and
May 2021. Patients were seen at irregular intervals. Their symptoms, the type and
duration of hormone intake, possible pregnancies and interventions, and DE dimensions
were documented. For data analysis, every visit was screened, but only those in which
there were changes compared to the previous visit were included in the study.
Transvaginal examination followed a standardized protocol. The typical transvaginal
sonographic pattern of rectal endometriosis is a hypoechogenic widening of the muscle
layer, visualized in the sagittal plane ([Fig. 1 ]) and documented photographically. The rectosigmoid was also inspected caudally and
cranially to the lesion to identify or exclude additional separate lesions. The measurement
included the length of the nodule, measured between the cranial and caudal poles of
the nodule, where the musculature pattern looks normal. The thickness measurement
was perpendicular to the length measurement and was taken at the widest part of the
lesion [13 ]
[14 ]
[15 ]. The individual measurements of nodule length and thickness for each patient over
time are depicted in [Fig. 2 ]. TVS was performed using the ultrasound device Sonoace SA-X8LV-GER (Samsung Medison
Co. Ltd.; Seoul, South Korea) with a 5.0–9.0 MHz transvaginal probe and the ultrasound
device Samsung WS80A (Samsung Medison Co. Ltd.; Seoul, South Korea) with a 5.0–9.0
MHz transvaginal probe.
Fig. 1 The sonographic image shows sample images of rectal endometriosis in three different
patients. Measurement of length (D1) and thickness (D2).
Fig. 2 These charts show each patient’s development of nodule length and thickness with increasing
age. Periods without hormone treatment are depicted with dashed lines.
The results were also classified using the Enzian and #Enzian classification, whereby
only the C-compartment was calculated for the study [16 ].
Statistical analysis
Data collection and management for this paper were performed using the OpenClinica
open-source software (Version 3.1, Copyright OpenClinica LLC and collaborators, Waltham,
MA, USA, www.OpenClinica.com ). The data set for this analysis can be found in the online supplement.
To model the temporal development of the length and thickness of the lesions, we fitted
mixed-effects models using the cumulative duration of hormone treatment (CDHT) up
to the considered visit and associated age as explanatory variables. To be more precise,
lesion size is assumed to be the sum of a random effect for each individual, a quadratic
function of CDHT, a cubic function of age, and a residual error for each observation.
The random effect allows us to adjust for the serial correlation of the measurements
within each patient and each patient’s starting size of the lesion. The choice of
a quadratic function in CDHT reflects the expectation that the longer the treatment,
the stronger its impact on lesion size, possibly non-linearly. The cubic function
can model the hypothesis that lesion size may initially increase with age and then
stagnate or even decrease. To interpret the fitted model, it is preferable to represent
the quadratic (cubic) function as a weighted sum of two positive quadratic (three
positive cubic) hat-shaped basis functions of CDHT (age), see panel A1 (A2) in supplementary fig. 1 . Technically speaking, the quadratic (cubic) functions in our models are represented
as B-splines of order 2 (3); see panel B1 (B2) in the supplementary fig. 1 and for more details, see [17 ]. We have also made the data set available in supplementary table 1 .
Data processing and statistical analysis were performed using the statistical programming
environment R, version 4.1.2 [18 ] and R-libraries lme4 and lmerTest to fit and evaluate random effects models.
Results
Our study included 38 premenopausal women with a mean age at first examination of
34.28 years. The average observation time was 7.24 years. Data points of an average
of 3.10 examinations per patient were considered during the observation period. Twenty
women had at least one previous gynecological operation before the first presentation
(rectal endometriosis operation excluded). Only four women had been successfully pregnant
before the primary presentation, and 18 were under hormonal treatment at the first
visit. During the observational period, 14 women underwent surgery (rectal endometriosis
operation excluded), 11 had been pregnant, and 15 were under hormonal treatment at
their last visit ([Table 1 ]).
Table 1 Characteristics of the study population. For numerical variables, the mean and standard
deviations are given. For classification, frequencies and, in brackets, relative frequencies
are shown.
Characteristics of the study population
Number of patients
38
Mean age at first visit in years
33.60 ± 6.21
Mean age at last visit in years
40.85 ± 5.96
Mean time of observation in years
7.24 ± 4.21
Mean number of included visits
3.10 ± 2.16
Women with a
previous gynecological operation
20 (52.6%)
Women with operations during the observational period
14 (36.8%)
Women with previous pregnancies
4 (10.5%)
Women who got pregnant during the observational period
11 (28.9%)
Hormone intake at first visit
18 (47.3%)
Hormone intake at last visit
15 (39.4%)
All women only had one rectal nodule detected sonographically. The average nodule
length at the first visit was 23.26 mm, and the average length at the last visit regressed
to 22.53 mm. The average nodule thickness at the first visit was 10.55 mm. Also, the
thickness regressed slightly at the last visit to an average of 8.81mm. Accordingly,
there were hardly any changes in the C-compartment of the #Enzian classification ([Table 2 ]).
Table 2 The characteristics of rectal endometriosis are shown. The values of nodule length
and thickness show the mean and the standard deviation. The #Enzian classification
classifies the rectal endometriosis into three grades depending on the length of the
nodules (C1: <1cm; C2: 1–3cm; C3: >3cm).
Characteristics of rectal endometriosis
Nodule length at first visit
23.26 ± 8.44 mm
Nodule length at last visit
22.53 ± 8.42 mm
Nodule thickness at first visit
10.55 ± 4.17 mm
Nodule thickness at last visit
8.81 ± 4.11 mm
#Enzian C compartment at first visit
C1
0
C2
29
C3
9
#Enzian C compartment at last visit
C1
1
C2
28
C3
9
To study the influence of age and duration of hormone treatment (CDHT) on lesion size,
we fitted mixed-effects models to our data. Since our data contain long-term observations,
it is to be expected that age and duration of hormone treatment will have a non-linear
effect on the temporal development of the lesion size.
As some women were successfully pregnant between two visits and subsequently started
hormone therapy again (and there was no visit during the pregnancy), we decided not
to include the pregnancy factor in our calculations to avoid possible misinterpretation.
Our model based on rectal nodule length has a significant negative correlation with
CDHT. Furthermore, there is a significant positive correlation with age, where an
increase in the length of the lesion until the 4th decade of life can be seen with
a stabilization effect of length afterwards ([Table 3 ]). Regarding the thickness of rectal endometriosis, only CDHT shows a significant
negative correlation. Age does not seem to influence nodule thickness ([Table 3 ]) significantly.
Table 3 The table shows the mixed effects models of the length and thickness of the rectal
nodule. For each patient, her random effect is considered as the realization of independent
normally distributed random variables with zero mean and unknown standard deviation
whose estimate is given in sub-tables a). These random effects may be considered as
the patient’s individual deviation from a common intercept. Similarly, the residual
errors are assumed to be normally distributed with zero mean and common standard deviation.
CDHT and age enter the model as the sum of two weighted sums of two and three B-spline
basis functions, respectively. The basis functions and the weighted sums are depicted
in the upper and lower row of Figure 1 in the Supplement, respectively. The weights are given in the “Estimate” column in
sub-tables b). For more details, see the section Statistical Analysis. An analysis
of the variance table in sub-tables c) examines the contribution of each B-spline
to the fit of the model. Significant results are marked with asterisks (p < 0.05:
*); SE: standard error, DF: degrees of freedom.
Nodule length
a) Random effects
Groups
Purpose
Variance
Estimate of standard deviation
Patients (n=38)
One random effect for each patient
71.441
8.452
Residual error (n=154)
One random error for each observation
9.957
3.156
b) Fixed effects
Basis function
Estimate
SE
DF
T-value
P-value
Intercept
15.9672
3.449
131.649
4.630
<0.0001*
B-spline (CTHT, degree = 2)
1st
0.3499
2.296
147.000
0.152
0.8791
2nd
–7.3109
3.129
137.063
–2.336
0.0209*
B-spline (age, degree = 3)
1st
10.9956
6.437
136.187
1.708
0.0899
2nd
9.5019
4.083
147.636
2.327
0.0213
3rd
8.1569
5.480
134.320
1.486
0.1397
c) Type III analysis of variance table with Satterthwaite’s method
Sum of squares
Mean squares
DF numerator
DF denominator
F-value
P-value
B-spline (CTHT, degree = 2)
67.805
33.902
2
118.27
3.405
0.0365 *
B-spline (age, degree = 3)
80.846
26.949
3
117.33
2.706
0.0485 *
Nodule thickness
a) Random effects
Groups
Purpose
Variance
Estimate of standard deviation
Patients (n=38)
One random effect for each patient
11.657
3.414
Residual error (n=154)
One random error for each observation
3.445
1.856
b) Fixed effects
Basis function
Estimate
SE
DF
T-value
P-value
Intercept
9.539
1.761
128.262
5.415
<0.0001*
B-spline (CTHT, degree = 2)
1st
–2.987
1.232
147.503
–2.426
0.0165*
2nd
–3.876
1.597
124.012
–2.426
0.0167*
B-spline (age, degree = 3)
1st
–0.464
3.587
145.233
–0.129
0.8972
2nd
3.469
2.166
144.323
1.602
0.1114
3rd
0.590
2.780
117.067
0.212
0.8323
c) Type III analysis of variance table with Satterthwaite’s method
Sum of squares
Mean squares
DF numerator
DF denominator
F-value
P-value
B-spline (CTHT, degree = 2)
32.343
16.172
2
109.27
4.695
0.0111*
B-spline (age, degree = 3)
9.058
3.010
3
107.42
0.877
0.4557
Discussion
Rectal endometriosis is a deeply infiltrating form of the disease. The origin and
the reason for the specific location of this form, and especially the time at which
the rectal lesion started, have not been clarified yet [4 ]. The extensive fibroblastic reaction with entrapped endometrial foci leads to marked
thickening of the rectal wall of varying length, thickness, and width. Some cases
show clinically relevant lumen narrowing and stiffening of the entire intestinal tube
[19 ]. The extent of findings plays an important role in evaluating symptoms and planning
and managing noninvasive and invasive treatments [20 ]
[21 ]. Transvaginal sonography and MRI of the pelvis have become the methods of choice
for imaging and measuring deep infiltrating endometriosis with high sensitivity and
specificity.
Since there is still little information about the onset of the disease and possible
factors influencing its growth or regression, both imaging techniques might help to
understand the growth dynamics of rectal endometriosis over time and the influence
of age and hormonal treatment. Fedele et al. could not detect any further growth trend
during the observation period using transrectal ultrasound [9 ].
In contrast, Netter et al., who monitored the TIE of the rectum via MRI, found partially
progressive nodules [1 ]. They could demonstrate that hormonal treatments, which induce amenorrhea, may prevent
the growth of nodules and may even result in the regression of lesions. However, the
regression or stability of nodule size did not correlate with pain relief in many
patients.
Barra et al. observed the regression of rectosigmoid endometriosis under Dienogest
therapy using a standardized ultrasound examination in a 3D model [10 ]. However, the observation period was maximum overall 36 months, and no findings
were recorded after cessation of therapy. A mean volume reduction of the nodules was
observed by at least 10%, but 5–10% of patients experienced an increase in volume
without worsening clinical symptoms. Knez et al. conducted a retrospective investigation
on a cohort of women with deep endometriosis who were not undergoing hormonal therapy
from various locations. The findings demonstrated that the number of endometriotic
nodules is a negative predictor for disease progression [22 ].
The effect of hormonal therapy on rectal endometriosis has been demonstrated by various
studies analyzing only the symptoms [12 ]
[23 ]. The use of the oral contraceptive pill, norethisterone acetate (NETA) [24 ], desogestrel, and triptorelin [25 ] decreases symptoms and improves quality of life. However, the relationship between
the size of the finding and the symptomatology is controversial. The available data
show partially divergent results regarding the growth behavior of the nodule in terms
of progression and regression, depending on the different affecting factors.
Our study aims to investigate the influence of hormone therapy and the age of the
patient on the growth and regression of the deep lesion in the rectal wall in a multifactorial
analysis. We observed the lesions intraindividually over a period of time, with the
longest period being more than 16 years.
To describe the growth pattern of the length and the thickness of the rectal foci,
we fitted multiple mixed-effects models, including the factors of age and hormone
intake (represented by the CDHT).
Our study shows that continuous administration of a progestogen or an estrogen/progestogen
medication in the long cycle reduces nodule size and thickness. As our models show,
there is no linear correlation between the regression of nodule length and thickness
and the duration of therapy. The histopathological composition of the nodule can explain
this. Most of the findings consist of fibrosis in which stroma and epithelial cells
are embedded. Fibroblasts are much less responsive to hormone modulation than stromal
and epithelial cells, which may explain the small reduction in size [26 ]. The fact that the relative length changes more than the thickness ([Fig. 3 ]) of the foci could be explained by mobility (fixation with surrounding structures)
and partly by the different dimensions. Proliferation may occur more in a longitudinal
direction than in a radial direction.
Fig. 3 Images A and D show the relative change in nodule length and thickness depending on age, as predicted
by our model as described in [Table 3 ] while assuming that there is no hormone intake. Images B and E show the predicted relative change in nodule length and thickness depending on the
cumulative time of hormone treatment (CDHT) so far, assuming the patient is 35 years
old. The gray bands indicate point-wise 95% confidence intervals. Images C and F show the growth behavior of the length and thickness of rectal endometriosis depending
on the duration of hormone intake. The age is shown in years. While the sample patient
on the black curve has never taken hormones (same curve as in images A and D ), the sample patient on the green curve has taken hormones continuously over a period
of 15 years. The patients shown in the red and blue curves started hormone therapy
at the age of 25 and paused therapy after 5 and 10 years, respectively.
The growth tendency of the rectal lesions, depending on the patient’s age, was significantly
increased between 20 and 40 years of age. Similar trends were found by Koninckx in
his retrospective study comparing the different focal sizes with age [27 ]. His data refer only to surgical and histological findings. The accuracy of the
measurements can be compared with the sonography findings to a limited extent.
The #Enzian classification within the C-compartment (<1 cm = C1, 1–3 cm = C2, >3 cm
= C3) can identify the changes in rectal foci size. However, the threshold values
for the individual C-compartments are probably too large to be able to calculate a
significance of the change in the sizes.
Our study has several limitations. Firstly, we have a study population in which a
primarily conservative approach was chosen, and no operative treatment of rectal endometriosis
was necessary until the end of the observation period. For instance, Roman et al.
described symptomatic and size-progressive rectal nodules that required rapid surgical
treatment [28 ]. The exclusion of patients with nodules exhibiting a more aggressive growth behavior
hinders the generalizability of the study results to all patients with deep infiltrating
rectal endometriosis. However, this limitation is a common issue, as the few other
studies investigating growth patterns over time have also focused on women without
prior rectal surgery [1 ]
[6 ]. Another limitation of our study is the rather small study population.
With correspondingly fewer data points from patients in their early 20s or late 40s,
this naturally leads to a larger confidence interval. Therefore, the curves should
not be overinterpreted in these specific ranges. However, it must also be acknowledged
that a single experienced gynecologist examined all patients. Thus, high accuracy
and precision of the measurements can be assumed. Additionally, our study was designed
retrospectively, and data was collected from patient records. Finally, there is a
known inter- and intra-observer variability of measurements using TVS, as observed
by Egekvist et al., which may influence the presented results [29 ]. Therefore, further studies covering the entire reproductive period are warranted
to gain a more comprehensive understanding of the growth behavior of rectal endometriosis.
Conclusion
Transvaginal sonography is an ideal method to study and monitor morphologic changes
of rectal endometriosis over time for clinical management considerations.
The growth pattern of deep endometriosis (DE) is influenced by patient age and the
duration of conservative therapy. In patients without any treatment, rectal endometriosis
shows a moderate increase in size until the end of the fourth decade of life, after
which it tends to stabilize. While hormonal therapy can reduce the size or prevent
the further progression of deep endometriosis, its growth does not follow a linear
function.