Expectant management of ultrasonically diagnosed ovarian dermoid cysts: is it possible to predict outcome?

Authors


Abstract

Objective

The aim of this study was to assess the natural history of ultrasonically diagnosed ovarian dermoid cysts in a large group of women who were managed expectantly, and to assess the factors that were associated with failure of expectant management.

Methods

Our database was searched for dermoid cysts diagnosed on ultrasonography by a single expert operator between 2001 and 2007 in this retrospective study. In women who opted for expectant management, demographic data including age, gravidity and parity were recorded. Indications for scan, site of cysts, dimensions and the outcomes of expectant management were also recorded.

Results

Two hundred and eighty-nine women were diagnosed with a total of 323 dermoid cysts by a single expert ultrasound operator. 93/289 (32.2%; 95% CI, 26.8–37.6%) women with 105/323 (32.5%; 95% CI, 27.4–37.6%) ovarian dermoid cysts were managed expectantly for longer than 3 months. The mean age at diagnosis was 33.8 (range, 13–79) years and the median duration of follow up was 12.6 (interquartile range, 7.6–29.3) months. The mean growth rate of dermoid cysts during follow up was 1.67 mm/year. There were no demographic or morphological features that could be used to predict the growth rate of dermoid cysts. After a period of expectant management, 24/93 (25.8%; 95% CI, 16.9–34.7%) women had surgical intervention. The risk of surgical intervention was significantly increased in younger women, those of parity ≥ 2 and in women with bilateral cysts or larger-diameter cysts, and reduced in women with a past history of ovarian cyst.

Conclusions

The success rate of expectant management of dermoid cysts is high and this approach should be considered as a viable alternative to surgical management. Copyright © 2010 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

Ultrasonography has increasingly been used for the assessment of women presenting with a wide range of gynecological complaints. Ultrasound examination involves a detailed assessment of the uterus and the ovaries, which consequently leads to some women being diagnosed with ovarian abnormalities that are not the cause of her presenting symptoms1. In these cases there is no compelling reason for surgical intervention, and expectant management is often seen as a preferred option.

Mature cystic teratomas or dermoid cysts are one of the most common benign ovarian tumors, with a reported prevalence of up to 20%2, 3. They occur bilaterally in 10–15% of cases3. Most dermoid cysts are diagnosed in women of reproductive age and are often discovered incidentally in asymptomatic women4. Possible complications associated with dermoid cysts include torsion, rupture and infection, although these complications are rare5, 6. In their pure form dermoid cysts are usually benign, however, malignant foci with squamous cell carcinoma are seen in approximately 1–2% of cases5, 7.

Once diagnosed, dermoid cysts have routinely been removed by elective surgery and this procedure accounts for 20–35% of all ovarian tumors removed surgically3, 8, 9. One reason for removing dermoid cysts in asymptomatic patients is concern regarding the increasing difficulty of surgery with growth of the cyst, the perceived increased risk of torsion and sometimes the risk of malignancy. However, there are very few data in the literature regarding the natural history of ovarian dermoid cysts10.

Transvaginal ultrasonography has an established role in the evaluation of adnexal masses, and provides an accurate assessment of ovarian morphology. It is also well known that dermoid cysts have distinct sonographic gray-scale morphological features, which can be used to accurately diagnose dermoid cysts (‘pattern recognition method’)11–15.

The aim of this study was to assess the natural history of ultrasonically diagnosed ovarian dermoid cysts in a large group of women who were managed expectantly, and to assess factors that were associated with failure of expectant management, thus necessitating surgical intervention.

Patients and Methods

This retrospective study was based at the Gynaecological Assessment Unit at King's College Hospital, London, UK, which is a tertiary referral gynecological ultrasound center. We identified all women who were examined by the same expert operator (D.J.) and who presented with typical morphological features of benign ovarian dermoid cysts on ultrasound. Diagnosis of dermoid cyst was based on the subjective assessment of ovarian morphology and pattern recognition technique. Pathognomonic features of dermoid cysts include the presence of a discrete highly echogenic focus with posterior shadowing (Rokitansky protuberance), fine echogenic bands (representing hair) within the cystic area and the presence of a fat–fluid level14–19. The site of the cyst was determined and the cyst was measured in three perpendicular planes. The surrounding tissue and ovarian capsule were not included in the measurements. The maximum diameter was recorded in all cases. The cyst's volume was calculated using the formula for an ellipsoid (4/3π abc, where a, b and c are semi-axes of the cyst).

Symptomatic women were offered surgical treatment, while asymptomatic women were given a choice between expectant management and surgery. Only women who opted for expectant management of dermoid cysts were included in the study. Women with less than 3 months follow-up were excluded from the study. In all cases the woman's age, gravidity, parity, indication for scan and presenting symptoms were recorded. Women who were referred for a scan with a previous history of an ovarian cyst managed either surgically or expectantly were grouped together as having a ‘past history of ovarian cyst’. The number of follow-up scans and the time intervals between the scans were also recorded. At each follow-up visit, every woman had a repeat ultrasound assessment, including systematic examination of the ultrasonographic features of the cyst. The size and appearance of the cyst were compared to the previous scan findings. In women who underwent surgical intervention after a period of expectant management the indication for surgery and time interval between presentation and intervention were recorded. The preoperative ultrasound findings were compared with the histology.

Statistical analysis

The statistical significance of age differences between women who were operated on and those who were not was evaluated with t-tests. The Wilcoxon rank-sum test was used to compare differences in maximum cyst diameters between these two groups. The change in volume per week of the dermoid cysts between the scans was assessed with the Kruskal–Wallis equality-of-populations rank test.

Predictive factors were sought from a multivariable logistic regression model with operation status as the outcome variable. Candidate factors included were age (scaled by 10-year increases), site (left, right, bilateral), number of pregnancies (categorical: 0, 1, 2, ≥ 3), parity (categorical: 0, 1, 2, ≥ 3), maximum cyst diameter at final scan and indication for scan (transient abdominal pain, mass/distension, past history of ovarian cysts and others). The categorical variables for number of pregnancies and parity were collapsed into four levels because there were very few cases of gravidity and parity greater than 3.

Robust standard errors were used to adjust for clustering within patients (i.e. two cysts per patient). A parsimonious model was sought by removing non-significant factors according to the Wald test, with the significance level set at 5%. Parameter estimates were expressed in terms of odds ratios (OR) and associated 95% CIs.

Goodness-of-fit diagnostics for the developed model included a Hosmer–Lemeshow test with 10 groups, deviance residual analysis and a receiver–operating characteristics (ROC) curve analysis of discrimination between observed and fitted values.

Predictive factors for cyst growth were sought from a mixed-effects linear regression model with individual patients accorded a random intercept20. The same candidate factors were assessed after adjustment for time. The diameter outcome was transformed with a square root to reduce the heterogeneity of growth rates. Statistical analysis was performed using Stata 10.1 software (StataCorp, College Station, TX, USA).

Results

Between 1 January 2000 and 1 January 2007, we identified 289 women with 323 ovarian dermoid cysts who were scanned by a single expert operator. 188/289 (65.1%; 95% CI, 59.6–70.6%) women opted for surgery, while 101/289 (34.9%; 95% CI, 29.5–40.5%) women with 113/323 (35.0%; 95% CI, 29.8–40.2%) ovarian dermoid cysts had expectant management and underwent a minimum of two scans in our department. Eight patients with eight ovarian dermoid cysts were excluded from the study because they had a follow-up period of less than 3 months. Therefore 93/289 (32.2%; 95% CI, 26.8–37.6%) women and 105/323 (32.5%; 95% CI, 27.4–37.6%) cysts were included in the study. The study population is illustrated in Figure 1. The mean age of the women was 33.8 (range, 13–79) years. 87/93 (93.5%) women were premenopausal and the remaining six (6.5%) women were postmenopausal. Unilateral cysts were found in 81/93 (87.1%; 95% CI, 80.3–93.9%) women, while 12/93 (12.9%; 95% CI, 6.1–19.7%) women had bilateral cysts. In women diagnosed with unilateral cysts, right-sided dermoid cysts were more common, though not statistically so, than unilateral left-sided ones (48/81 (59.3%; 95% CI, 48.6–70.0%) vs. 33/81 (40.7%; 95% CI, 30.0–51.4%); P = 0.23).

Figure 1.

Flow chart illustrating the study population.

At presentation, the majority of women (65/93 (69.9%; 95% CI, 60.6–79.2%)) who opted for expectant management were asymptomatic, 18/93 (19.4%; 95% CI, 11.3–27.4%) women complained of transient mild abdominal pain and 10/93 (10.8%; 95% CI, 4.5–17.1%) women were referred following a clinical diagnosis of pelvic mass or swelling (Table 1). 17/93 (18.3%; 95% CI, 10.4–26.1%) women were pregnant at the time of diagnosis. The median duration of follow-up was 12.6 (interquartile range (IQR), 7.6–29.3) months, with a range of 3–73 months. The maximum cyst diameter increased by a median of 0.09 (IQR, − 0.09 to 0.52) mm/month and the median volume increase was 0.09 (IQR, − 0.013 to 0.871) mL/month. No serious complication such as torsion, rupture or infection was noted in our patients during the follow-up period.

Table 1. Clinical presentation and management of women with ovarian dermoid cysts and indications for surgical intervention
Parametern% (95% CI)
Clinical presentation  
 Transient abdominal pain18/9319.4 (11.3–27.4)
 Clinical diagnosis of pelvic mass10/9310.8 (4.5–17.1)
 Incidental finding in non-pregnant48/9351.6 (41.5–61.8)
  woman  
 Incidental finding in pregnancy17/9318.3 (10.4–26.1)
Management  
 Expectant69/9374.2 (65.3–83.1)
 Surgical intervention24/9325.8 (16.9–34.7)
 Indication for surgery  
  Patient request12/2450.0 (30.0–70.0)
  Symptomatic (abdominal9/2437.5 (18.1–56.9)
   pain/bloating)  
  Other3/2412.5 (0–25.7)

Factors for predicting growth rate

Mixed-effects linear regression did not show any factors that could predict the growth rate of dermoid cysts. Model diagnostics also did not fit well and the normality of the residuals deteriorated considerably when the non-significant factors were removed from the model. Other transformations of the diameter outcome were attempted (logarithm and cube root) but were found to worsen the fit.

Factors for predicting risk of surgical intervention

After a period of expectant management, 24/93 (25.8%; 95% CI, 16.9–34.7%) women had surgical intervention to remove their dermoid cysts. In all cases, the diagnosis of benign dermoid cysts was confirmed by histology. The mean age of women who had surgical intervention was 29.8 ± 8.2 years at the time of initial diagnosis, which was significantly lower than 35.0 ± 10.3 years in women who had successful expectant management (t(66.9) = 2.72, P = 0.008). The maximum diameters of dermoid cysts at initial scans were significantly different between women with successful and failed expectant management (median 29 (IQR, 21–47) mm vs. 57 (IQR, 41–75) mm; Z = − 4.31, P < 0.001). The maximum diameter of dermoid cysts measured just before surgery was 74 (IQR, 50–89) mm, which was significantly higher than the last recorded maximum diameter of women with successful expectant management (34 (IQR, 22–47) mm; Z = − 5.46, P < 0.001).

Twelve out of 24 surgical interventions (50.0%; 95% CI, 30.0–70.0%) were performed at the patient's request. In total, 9/24 (37.5%; 95% CI, 18.1–56.9%) women became symptomatic with complaints of pain or abdominal distension after a period of expectant management and were operated on. Three out of 24 (12.5%; 95% CI, − 0.7 to 25.7%) women had ovarian cystectomy during an unrelated operation (Table 1). Of all surgical interventions, 18/24 (75.0%; 95% CI, 57.7–92.3%) women had laparoscopies while the remaining six (25%; 95% CI, 7.7–42.3%) had laparotomies. Overall, 22 (91.7%; 95% CI, 80.61–100.0%) women had ovarian cystectomies while two (8.3%; 95% CI, 0–19.4%) women had oophorectomies. The median volume of the dermoid cysts at the final scan in the group that had elective surgery was 83.5 (IQR, 49.0–119.8) mL compared with 217.5 (IQR, 71.3–257.0) mL in the group that developed symptoms (Z = 1.74, P = 0.08). The median change in volume in the two groups was 4.1 (IQR, − 1.5 to 65.5) mL and 78.5 (IQR, 4.0–197.0) mL, respectively. There was a large difference in the median change of volume between the two groups of patients from initial presentation to final scan, but this was not statistically significant (Z = 1.89, P = 0.06).

Of the 17 pregnant women with dermoid cysts, nine (52.9%; 95% CI, 29.2–76.7%) eventually had their cysts removed postpartum. Three women had complaints of pain with the dermoid cysts during pregnancy but none of them required surgery during pregnancy. No significant change in the size of the dermoid cysts was noted during pregnancy.

The majority of intervention occurred during the first 18 months of follow-up. A graph of the cumulative duration of expectant management prior to surgical intervention (Figure 2) shows that by 6 months, 9/24 (37.5%; 95% CI, 18.1–56.9%) women had surgical removal of their dermoid cysts. At 1 year, this had increased to 16 (66.7%; 95% CI, 47.8–85.5%) women and after 2 years, 20 (83.3%; 95% CI, 68.4–98.2%) women had had surgical interventions for the ovarian dermoid cysts.

Figure 2.

Cumulative operation rate by duration of expectant management prior to surgical intervention.

Multivariable logistic regression analysis showed five significant independent prognostic variables (Table 2) associated with surgical intervention (Pseudo R2 = 0.4622). A 10-year increase in age decreased the odds of an operation by 0.24 times (95% CI, 0.09–0.65) (Figure 3). Multiparous women were found to have significantly increased odds of surgical intervention when compared to nulliparous women. Parity of 2 increased the odds of an operation by 13.12 times (95% CI, 1.86–92.53), while parity of ≥ 3 increased the odds by 12.15 times (95% CI, 1.31–112.35) relative to nulliparity. Patients who presented for the initial scan with a past history of ovarian cyst were 0.10 times (or 90% less) likely to have surgical intervention, relative to patients who had transient abdominal pain (P = 0.041). Bilateral ovarian dermoid cysts increased the odds of an operation by 18.39 (95% CI, 1.75–193.38) times relative to a unilateral left ovarian dermoid cyst. When bilateral cysts were compared against unilateral right cysts there was a similar trend, which was not statistically significant (Wald χ2(1) = 3.63, P = 0.057). Finally, as expected, there was a significant relationship between cyst diameter and the risk of surgical intervention. Patients with larger cysts were more likely to have their cysts removed (Figure 4)—a 1-mm increase in final cyst diameter increased the odds of surgery by 1.05 (95% CI, 1.01–1.09) times. A Hosmer–Lemeshow goodness-of-fit test with 10 groups suggested that the model fitted well (χ2(8) = 4.22, P = 0.837). The area under the ROC curve (AUC) suggested good discrimination performance between the observed and predicted outcome (AUC = 0.93; exact binomial 95% CI, 0.87–0.97).

Figure 3.

Predicted probability (_____), calculated using multivariable logistic regression analysis, of surgical intervention according to age in patients who had opted for expectant management of ovarian dermoid cysts. Shaded area shows 95% CI.

Figure 4.

Predicted probability (_____), calculated using multivariable logistic regression analysis, of surgical intervention according to cyst diameter at final scan in patients who had opted for expectant management of ovarian dermoid cysts. Shaded area shows 95% CI.

Table 2. Results of multivariable logistic regression analysis showing factors influencing outcome of expectant management in 93 women with 105 ovarian dermoid cysts
ParameterOdds ratioRobust SE (95% CI)*P
  • Log pseudolikelihood, − 33.79. McFadden's Pseudo R2, 0.46.

  • *

    Based on robust sandwich estimator and 100-patient clusters.

  • Bilateral vs. right was non-significant (95% CI, 0.94–83.78, P = 0.057).

Age (10-year linear increase)0.240.12 (0.09–0.65)0.005
Parity   
 13.733.20 (0.69–20.05)0.126
 213.1213.07 (1.86–92.53)0.010
 ≥ 312.1513.79 (1.31–112.35)0.028
Indication for scan   
 Mass/distension0.550.59 (0.07–4.53)0.580
 Past history of ovarian cyst0.100.11 (0.01–0.91)0.041
 Other0.910.75 (0.18–4.56)0.908
Site   
 Right vs. left2.071.45 (0.53–8.17)0.298
 Bilateral vs. left18.3922.08 (1.75–193.38)0.015
Maximum cyst diameter at final scan1.050.02 (1.01–1.09)0.012

Discussion

Traditionally, women with complex ovarian cysts such as dermoid cysts have been routinely offered surgical treatment because of uncertainty in the nature of the lesion. Advances in the quality of ultrasound equipment and better experience of ultrasound operators have resulted in an increased accuracy of ultrasound diagnosis of dermoid cysts. Although in some cases, morphological features of dermoid cysts are unusual, typical ultrasound features are almost pathognomonic of a benign dermoid cyst. This increased diagnostic confidence eliminates the fear of cancer as the primary indication for surgery. In asymptomatic women with a reliable diagnosis of dermoid cyst, the possible benefits of surgical intervention are not clear and the risks and costs of surgery are hard to justify. In our study, one third of women with ultrasonically diagnosed ovarian dermoid cysts were managed expectantly. However, the majority of women with dermoid cysts were asymptomatic or had minimal clinical symptoms, which indicates that the proportion of women who could be offered expectant management is likely to be higher in the future with a better understanding of the benefits and risks of expectant management.

The choice between expectant management and surgical treatment at initial presentation is likely to be heavily influenced by the opinions of the attending clinicians. However, once women opt for expectant management, it is the clinical symptoms rather than preference of the clinician that will determine the need for surgical intervention. In women who selected expectant management, the success rate was 74.2% after a median follow-up of 12.6 months. This would increase to 77.4% if we eliminate women who had opportunistic cystectomy during surgery for unrelated indications. Surgical interventions were mainly performed at the women's request (50.0%) or because of clinical symptoms such as abdominal pain or distension (37.5%). We found five factors that significantly affected the risk of surgical intervention: age, parity ≥ 2, past history of ovarian cyst, bilateral cysts and larger cyst size.

The increased risk of surgical intervention in younger women may reflect concerns regarding possible adverse effects of cysts on their future fertility. Younger women may also feel that owing to their longer life expectancy, repeated follow-up visits may become tiresome and that there is a longer time span for cyst-related complications to develop. Younger women tend to be in better health and they may be less concerned about the effects of surgery on their general health than older women.

Multiparous women with parity of greater than two had at least 12 times greater odds of surgical intervention compared to nulliparous women. Their tendency to choose surgery may be influenced by the social pressures of looking after large families and their reluctance to accept the risk of an unplanned emergency operation. Multiparous women may also be less concerned about the possible loss of their ovaries during surgery, which may also affect their management decision.

The increased risk of failed expectant management in bilateral dermoid cysts may be caused by the increased risk of cysts becoming symptomatic owing to the combined volume effects. The presence of bilateral cysts may also be interpreted by women as having twice the risk of cyst complications and potential loss of both ovaries owing to possible torsion.

Larger dermoid cysts increase the risk of surgical intervention. It is possible that the sheer knowledge of a larger cyst contributes to the patient's perception of cyst symptoms, leading to increased surgical intervention. Medical professionals also tend to take the cyst's diameter into account when advising on the management of ovarian cysts, which may have contributed to the increased intervention rates in larger cysts.

In view of these findings, the assessment of dermoid cyst growth rate is important and it may be helpful in assessing the risk of surgical intervention in cysts managed expectantly. In a previous study involving 72 premenopausal women with ultrasound diagnosis of dermoid cyst, Caspi et al.10 reported a mean cyst growth rate of 1.8 mm/year. Another study, by Alcazar et al.21, which includes 17 dermoid cysts managed expectantly, also showed that dermoid cysts have a tendency to grow slowly over a period of time. To compare our study with previously published studies, we calculated a mean diameter growth rate of 1.67 mm/year, which is very similar to the findings of the previous studies. Such a slow growth rate implies that it takes an average of 6 years for the maximum diameter of an ovarian dermoid cyst to increase by 1 cm. However, the tendency for dermoid cysts to grow continuously may be used to justify surgery in younger women, while expectant management could be more appropriate in older women for whom operative risks are increased and where the cysts are unlikely to grow substantially by the time the women reaches her mean life expectancy age.

Dermoid cysts, similar to uterine fibroids, are commonly detected during routine ultrasound surveillance of the pelvis. In the majority of cases they are not the cause of the patient's symptoms and yet their presence is often considered an indication for surgery. In recent years there has been an emerging consensus that uterine fibroids should not be operated on unless they are causing clinical symptoms. In the case of dermoid cysts, the operation is advised either because of uncertainty about the nature of the cyst or because of concerns about the risk of complications occurring in the future. The only potential benefit of early intervention in this instance would be a high risk of complications occurring later in women in whom the operation is delayed. The results of our study, as well as those of previous reports, show that the risk of complications in women who choose expectant management is low, which may help in counseling women with dermoid cysts about available management options. The true potential benefit of surgical intervention in asymptomatic dermoid cysts could only be assessed in a prospective randomized trial. However, such a study is unlikely to occur because women tend to have strong preferences about their management. In addition, adverse outcomes with surgery are relatively infrequent and long-term complications such as infertility or pelvic pain are difficult to assess accurately.

In summary, our study has shown that expectant management of ovarian dermoid cysts is feasible in women in whom the diagnosis is made incidentally on ultrasound scan. The risk of cysts becoming symptomatic is low, but intervention should be considered in younger multiparous women, particularly if the cysts are large and bilateral, as expectant management is least successful in these cases.

Ancillary