Recurrence rate and associated factors of borderline ovarian tumors in the south of Vietnam

Abstract

Aim:

This study aimed to determine the recurrence rate and related risk factors of borderline ovarian tumors (BOT).

Methods:

We conducted a retrospective cohort study with 433 patients who were surgically treated for primary BOT at Tu Du Hospital from 11/2008 to 09/2015. We used the life table method to estimate the cumulative recurrence rate. We used the log–rank test and Cox proportional hazard model to determine recurrence-associated factors.

Results:

Median follow-up time was 43 months (range: 3–105 months). Eighteen patients developed recurrence. The cumulative BOT recurrence rates at year 1, 2, 3 and 4 were 1.2% (95% confidence interval [CI] = 0.5–2.8), 3.0% (95% CI = 1.7–5.2), 4.6% (95% CI = 2.9–7.4), and 5.1% (95% CI = 3.2–8.0), respectively. In the final multivariate model, a higher recurrence rate was significantly associated with primary tumor stages (stage I vs stages II and III, hazards ratio [HR] = 4.44, 95% CI = 1.60–12.38), pre-operative tumor’s capsule rupture (HR = 4.14, 95% CI = 1.78–9.64), and cystectomy (HR = 5.33, 95% CI = 1.43–19.91).

Conclusion:

The overall BOT recurrence rate in women in southern Vietnam was moderate. Primary tumor stage, capsule rupture, and cystectomy were main factors associated with BOT recurrence. Appropriate follow-up strategies for patients with high-risk factors are needed for early detection and management of recurrence.

Introduction

Borderline ovarian tumors (BOT) are ovarian tumors of epithelial origin with proliferation of the stratified epithelial lining layer, but without stromal invasion.¹ BOT account for 15–20% of malignant tumors originating from the epithelium. In western countries, the BOT incidence rate is 1.8–4.8 per 100 000 women per year; incidence has increased in recent decades.¹ The 5- and 10-year survival rates for BOT are 95.0% and 92.8%, respectively, and recurrence rates range from 5% to 8%.^2–4 On average, patients with BOT are 10 years younger than those with ovarian cancer.⁵

Borderline ovarian tumors have two main histological types: serous and mucinous.⁶ Others types include endometrioid, clear cell, Brenner and mixed types. The main treatments for BOT are radical or conservative surgery. The conservative approach, which preserves the uterus and some ovarian tissues, is used for young women who desire children. At present, there is no evidence that chemotherapy reduces the recurrence rate or improves survival in BOT patients. While BOT have better prognosis than do other ovarian cancers, approximately 10–15% of patients treated for BOT will relapse or develop a clinical status of invasive epithelial cancer.⁷ Most cases of BOT that recur will relapse within the first 2 years.⁸

Examining the risk factors of BOT recurrence is important for several reasons, including prediction of prognosis, post-operative monitoring for patients at risk, appropriate counseling for patients, and developing preventive treatments for recurrence. Still, factors related to BOT recurrence remain controversial. In southeast Asia, especially in Vietnam, there have been few studies of recurrent BOT.

Located in Ho Chi Minh City, Tu Du Hospital (TDH) is the largest, national-level obstetrics/gynecology hospital in the south of Vietnam. Each year, TDH provides treatment and cares for more than 500 000 patients from 32 southern provinces of Vietnam. At TDH, there are approximately 70 BOT cases per year, out of 240 cases of ovarian cancer. The present study aimed to determine the recurrence rate and related risk factors of BOT at TDH.

Methods

We employed a retrospective cohort study design. Medical records of all patients diagnosed with BOT at TDH from 11/2008 to 09/2015 were reviewed. We included all BOT cases that had anatomo-pathologic confirmation of diagnosis at TDH. Exclusion criteria were: (i) having other cancers, (ii) having acute or chronic illnesses that affected patients’ life span, and (iii) incomplete or missing medical records. The study was approved by the Institutional Review Board of the TDH (No.: 1649/QD-BVTD/Oct 6th, 2016).

Data obtained from the medical records included basic demographic information, BOT characteristics, BOT treatment and BOT relapse. Tumor stage classification was based on to the International Federation of Gynecology and Obstetrics (FIGO) staging system for ovarian carcinoma.² Recurrent BOT was defined as cases that had anatomopathologically confirmed primary BOT and were surgically treated, discharged with stable clinical outcomes, and then experienced one of the following: (i) a recurrent increase in CA-125 level according to Rustin’s standards,⁹ (ii) a recurrent BOT confirmed by clinical examination or diagnostic imaging, or (iii) a recurrent BOT confirmed by anatomopathology if the recurrence was surgically treated. Rustin’s standards for recurrence based on CA-125 concentration are: (i) CA-125 > 2 times the upper normal limit on two occasions at least 1 week apart (in patients with CA-125 in the normal range before treatment for primary BOT, or in patients with elevated CA-125 pretreatment that returned to normal after treatment for primary BOT); or (ii) CA125 > 2 times the nadir value on two occasions at least 1 week apart (in patients with elevated CA-125 before treatment for primary BOT that never normalizes).¹⁰ At TDH, after treatment for primary BOT, patients were followed once a month for the first 6 months, once every 2 months for the next 12 months, once every 3 months for the next 18 months, once every 6 months for the next 24 months (i.e., the end of 5 years after treatment), and then once a year.

We analyzed the data with Stata 13.0 (StataCorp). Time-to-recurrence (in months) was defined as the time from treatment completion of primary BOT to the time of diagnosed recurrence. Overall follow-up time for non-recurrences was from the time of treatment completion for primary BOT to the time of last follow-up, end of study, or death. Descriptive analyses with percentages and medians were performed to examine patients’ and tumors’ characteristics. We used the life table method to estimate the cumulative recurrence rate. To examine the relationship between categorical characteristics and recurrence time in univariate analyses, we used the log–rank test of equality. For multivariate analyses, we used Cox proportional hazard regression. Variables included in multivariate models were selected based on bivariate associations (P-value < 0.25) and a priori knowledge. The final multivariate model was chosen based on Akaike Information Criteria (AIC).

Results

Among 433 BOT patients included in the analyses from 11/2008 to 12/2016, one patient died due to the primary BOT, three patients died due to non-cancer causes and 18 patients were diagnosed with recurrent BOT. Patients’ and tumors’ characteristics are described in Table 1. Patient ages ranged from 16 to 76 years, with a median age of 38 years. More than half (60.7%) had ≥1 children. Most primary BOT were classified as mucinous (72.1%) or serous (24.2%). CA125 levels >35 U/mL were found in 36.6% of confirmed BOT. Of these cases with elevated CA125, 35.5% were serous tumor type and 55.5% were mucinous tumor type. The majority of primary BOT were at FIGO stage I (89.4%). Of all primary BOT, 56.8% had pre-operative capsule rupture, 56.8% had tumor sizes ≥10 cm, and 91.9% had bilateral BOT. In the preliminary bivariate analyses using proportions, recurrent BOT was statistically associated with pre-operative CA-125 level, clinical stage, peritoneal implants, tumor’s capsule rupture, tumor location, surgical procedure and chemotherapy administration.

Table 1.

Clinical and histopathologic characteristics of patients with borderline ovarian tumors (BOT)

Characteristics	Total	No recurrence (N = 415)	Recurrence (N = 18)	P-value*
	n (%)	n (%)	n (%)
Age				0.179
16–39 years old	241 (55.7)	234 (54.0)	7 (1.6)
≥ 40 years old	192 (44.3)	181 (41.8)	11 (2.6)
Nulliparous				0.951
Yes	170 (39.3)	163 (37.6)	7 (1.6)
No	263 (60.7)	252 (58.2)	11 (2.6)
Pre-operative CA-125 level				0.004
≤35 U/mL	269 (63.4)	263 (62.0)	6 (1.4)
>35 U/mL	155 (36.6)	143 (33.7)	12 (2.8)
Histological type				0.654**
Mucinous	312 (72.1)	300 (69.3)	12 (2.8)
Serous	105 (24.2)	100 (23.1)	5 (1.1)
Endometrioid	13 (3.0)	13 (2.5)	0 (0.0)
Clear-cell	1 (0.2)	1 (0.5)	0 (0.0)
Mixed	2 (0.5)	2 (0.5)	1 (0.2)
Clinical stage				<0.001***
Ia and Ib	195 (45.0)	195 (45.0)	0 (0.0)
Ic	192 (44.4)	184 (42.5)	8 (1.9)
II and III	46 (10.6)	36 (8.3)	10 (2.3)
Peritoneal implants				<0.001
Not invasive	431 (99.5)	414 (95.6)	17 (4.0)
Invasive	2 (0.5)	1 (0.2)	1 (0.2)
Pre-operative capsule rupture of the primary tumor				0.024
No	187 (43.2)	184 (42.5)	3 (0.7)
Yes	246 (56.8)	231 (53.3)	11 (2.5)
Tumor size				0.778
<10 cm	187 (43.2)	180 (41.6)	7 (1.6)
≥10 cm	246 (56.8)	235(54.3)	11(2.5)
BOT location				0.002
One side	398 (91.9)	385 (88.9)	13 (3.0)
Both sides	35 (8.1)	30 (6.9)	5 (1.2)
Surgical procedure				0.002
Conservative	210 (48.5)	208 (48.0)	2 (0.5)
Radical	223 (51.5)	207 (47.8)	16 (3.7)
Surgical type				<0.001
Unilateral cystectomy	14 (3.2)	13 (3.0)	1 (0.2)
Bilateral cystectomy	2 (0.5)	2 (0.5)	0 (0.0)
Unilateral cystectomy and salphingoo-oopherectomy	5 (1.1)	3 (0.7)	2 (0.5)
Unilateral salphingoo-oopherectomy	203 (46.9)	198 (45.7)	5 (1.2)
Bilateral salphingoo-oopherectomy	2 (0.5)	1 (0.2)	1 (0.2)
Hysterectomy plus bilateral salphingoo-oopherectomy	207 (47.8)	199 (45.7)	10 (2.1)
Surgical technique				0.315
Laparoscopy	115 (26.6)	112 (25.9)	3 (0.7)
Laparotomy	318 (73.4)	303 (70.0)	15 (3.5)
Cystectomy				0.008
No	412 (95.2)	397 (91.7)	15 (3.5)
Yes	21 (4.8)	18 (4.1)	3 (0.7)
Intra-operative frozen-section analysis (IFS)				0.918
No	365 (84.3)	349 (80.6)	16 (3.7)
Yes	68 (15.7)	66 (15.2)	2 (0.5)
IFS’s results				0.345
Benign	12 (17.7)	11 (16.2)	1 (1.5)
Malignant BOT	56 (82.4)	55 (80.8)	1 (1.5)
Chemotherapy				0.005
No	250 (57.7)	246 (56.8)	4 (0.9)
Yes	183 (42.3)	169 (39.0)	14 (3.3)

^*P-value from Log–rank test results;

^**P-value from Log–rank test for histological types with two values: mucinous vs. serous;

^***P-value from Log–rank test for clinical stages with two values: I vs. II and III.

Patients were followed for a median of 43 months (range: 3–105 months). The cumulative BOT recurrence rates after 1, 2, 3 and 4 years were 1.2%, 3.0%, 4.6% and 5.1%, respectively (Table 2). The shortest relapse duration was 4 months, the longest was 47 months, and the median was 18 months. The overall incidence rate was 0.09 person-months (95% confidence interval [CI] = 0.06–0.14). In the multivariate Cox regression model (Table 3), a higher recurrence rate was significantly associated with stage (stage I vs stages II and III), pre-operative capsule rupture of the primary tumor, and cystectomy. Patients with primary BOT at stages II and III were four times more likely to have recurrent BOT (hazards ratio [HR] = 4.44, 95% CI = 1.60–12.38), than were patients with stage I BOT. Similarly, patients with pre-operative capsule rupture of the primary tumor had a four-fold increase in recurrence risk (HR = 4.14, 95% CI = 1.78–9.64). In the subgroup analyses, recurrence risk of serous tumor type was related to clinical stage (P-value = 0.001), tumor’s capsule rupture (P-value = 0.003), and peritoneal implants (P-value<0.001) (data not shown). The risk of recurrence of stage I BOT was associated with tumor’s capsule rupture (P-value = 0.037) and peritoneal implants (P-value = 0.001) (data not shown).

Table 2.

Cumulative risk of having recurrent borderline ovarian tumors (BOT)

Month interval	Number of women without BOT recurrence entering the interval	Number of recurrent BOT	Number of censored cases	Estimated cumulative risk of having recurrent BOT % (95% CI)
0–12	433	5	13	1.2 (0.5–2.8)
12–24	415	7	67	3.0 (1.7–5.2)
24–36	341	5	92	4.6 (2.9–7.4)
36–48	244	1	47	5.1 (3.2–8.0)
48–60	196	0	44	5.1 (3.2–8.0)
60–72	152	0	50	5.1 (3.2–8.0)
72–84	102	0	40	5.1 (3.2–8.0)
84–96	62	0	41	5.1 (3.2–8.0)
96–108	21	0	21	5.1 (3.2–8.0)

Table 3.

Risk factors for recurrent borderline ovarian tumors (BOT)

Factors	Risk time (month)	Recurrence (N = 18)		Cox proportional hazards regression: Hazard Ratio (95% CI)
		n/Total	Incidence rate (person-1000 months)	Univariate analyses	Multivariate final model A*	Multivariate final model B*
Age
16–39 years	7766	7/241	0.64	1	-	-
≥40 years	12 973	11/192	1.13	1.89 (0.73–4.88)
P-value				0.187
Pre-operative CA-125 level
≤35 U/mL	13 475	6/269	0.45	1	1	1
>35 U/mL	6976	12/155	1.72	3.84 (1.44–10.24)	2.56 (0.90–7.24)	2.08 (0.74–5.83)
P-value				0.007	0.077	0.162
Histology
Serous	15 079	12/312	0.80	1	-	-
Mucinous	4852	5/105	1.03	1.27 (0.45–3.60)
P-value				0.655
Stage
I	18 429	8/387	0.43	1	1	1
II and III	2310	10/46	4.33	10.52 (4.15–26.67)	4.44 (1.60–12.38)	7.29 (2.73–19.44)
P-value				<0.001	0.004	<0.001
Peritoneal implants
Non-invasive	20 719	17/431	0.82	1	-	-
Invasive	20	1/2	50.0	37.56 (4.58–308.40)
P-value				0.001
Pre-operative capsule rupture of the primary tumor
No	9470	3/187	0.32	1	1	1
Yes	11 269	15/246	1.33	4.04 (1.79–9.11)	4.14 (1.78–9.64)	4.75 (1.77–12.76)
P-value				0.001	0.001	0.002
BOT location
One side	19 167	13/398	0.82	1	1	1
Both sides	1572	5/35	0.90	4.34 (1.55–12.18)	2.90 (0.91–9.28)	2.68 (0.89–8.06)
P-value				0.005	0.072	0.079
Surgical procedure
Conservative	9287	2/210	0.68	1	1
Radical	11 452	16/223	3.18	7.10 (1.63–30.90)	3.36 (0.71–16.05)
P-value				0.009	0.128
Cystectomy
No	16 656	15/412	0.001	1		1
Yes	690	3/21	0.004	4.56 (1.32–15.76)		5.33 (1.43–19.91)
P-value				0.016		0.013
Chemotherapy
No	11 735	4/250	0.34	1	-	-
Yes	9004	14/183	1.55	4.36 (1.43–13.24)
P-value				0.010
Surgical technique
Laparoscopy	5623	3/115	0.53	1	-	-
Laparotomy	15 116	15/318	0.99	1.87 (0.54–6.45)
P-value				0.324

^*The final model had the lowest value of the Akaike information criterion (i.e., the best model), compared with other models.

Discussion

We found that the overall BOT recurrence rate in women in southern Vietnam was 5.1%. This rate was within the BOT recurrence rate range of 3–10% reported by studies in other countries, such as 10% in Sweden,¹¹ 3.0% in Turkey¹² and 5.2% in northern Vietnam.¹³

Factors consistently associated with recurrent BOT in our adjusted analysis included stage and pre-operative capsule rupture of the primary BOT. Some previous studies have shown that higher FIGO clinical stages were associated with higher BOT recurrence rates.^8,11,14–17 For example, in a study of 339 Italian women, those who had primary BOT at a more advanced stage had a higher recurrence rate (40%) after conservative surgeries, compared with stage I (15%).¹⁸ Also, the associations between tumor’s characteristics (including rupture and stage) have been observed commonly in other studies¹⁶; thus, our results are consistent with these previous findings.

Although CA-125 concentration in serum is frequently used as an indicator for malignant ovarian tumors, it is not useful in BOT diagnosis and follow-up.^19–21 An increased level of CA-125 was found in 75% of serous BOT cases, but only in 30% of mucinous BOT cases.²¹ The average CA-125 absolute value of serous tumors was also higher than that of mucinous tumors. In our study, pre-operative CA-125 values of more than 35 U/mL were observed in 36.6% of cases (35.5% in serous BOT vs 55.5% in mucinous BOT). In the adjusted multivariate analysis, however, CA-125 level was not an independent prognostic factor of BOT recurrence (P-value = 0.077), although this result might be due to a small number of cases in each type subgroup.

Several researchers have suggested that a conservative surgical approach increased the risk of BOT recurrence.^16,22 Nevertheless, the relationship between surgical procedure and recurrence is influenced or mediated by other factors, such as staging, histological types, cystectomy and the type of surgical approach (laparoscopy vs laparotomy).¹⁶ In particular, the use of conservative surgery in advanced-stage BOT increased the risk of recurrence by 3–5 times compared with the use of radical surgery.^16,22 When controlling for stages of BOT in the multivariate model, we found that conservative or radical surgical approach was not related to BOT recurrence. Nevertheless, when we particularly examined whether the surgical type was cystectomy or not (Model B), cystectomy increased the risk of recurrence. This finding is consistent with results from a multi-center study, systematic reviews and meta-analysis which showed that cystectomy was a significant risk factor for BOT recurrence.^23–25

In the univariate analyses, peritoneal implants and chemotherapy were associated with BOT recurrence. These variables were included in initial multivariate models. However, based on AIC and a priori knowledge, we excluded these two variables from the final model. Whether peritoneal implants are a risk factor for BOT recurrence remains controversial. Some researchers found that the recurrence was associated with histological subtypes of peritoneal implants (invasive or non-invasive),^8,15,16 while others reported no difference in recurrence rates.²⁶ In our study, only two patients with treated primary BOT had invasive peritoneal implants, which might explain the inferior AIC value when this variable was initially included.

Several studies have shown that supplemental chemotherapy or radiotherapy did not have benefit in BOT treatment, even in advanced-stage BOT or BOT with invasive peritoneal implants.^27,28 Chemotherapy and radiotherapy may have a negative effect on BOT prognosis and patients’ overall health due to toxicity and complications. In addition, there is no clear evidence that chemotherapy can reduce BOT recurrence.²⁸ Given the unestablished efficacy of chemotherapy and a poor model fit when this variable was included, we excluded chemotherapy from the final model.

In our study, only 16% of patients underwent intraoperative frozen-section analysis (IFS) and the IFS results were not associated with BOT recurrence. Given the scope of this study, we did not examine further the sensitivity and specificity of IFS results. However, findings from a systematic review suggested that when IFS result indicated a borderline tumor, approximately 21% would be diagnosed as invasive epithelial ovarian cancer in the final pathologic assessment.²⁹ Therefore, extra caution is recommended when interpreting IFS results.

Our study had some limitations. First, it had a retrospective cohort design. Prospective studies with multiple participating centers in Vietnam or in neighboring countries are needed to generate stronger evidence regarding risk factors of recurrent BOT in the region. Second, also due to the retrospective nature of the study, several variables about patients’ characteristics and behaviors were not available for analyses. Third, although our cohort sample size was 433 patients, only 18 developed BOT recurrence, limiting statistical power for subgroup analyses. Finally, some patients might have had BOT recurrence, but they might have not realized it or might not have come to TDH for follow-up treatment as instructed.

In conclusion, our study found that the overall BOT recurrence rate in women in southern Vietnam was moderate (5.1%). Factors associated with BOT recurrence included primary tumor’s stage, pre-operative capsule rupture, and cystectomy. Appropriate follow-up strategies (e.g., more regular follow-ups up to 5 years) for patients with high-risk factors are needed for early detection and management of recurrence.