Unexpected malignancy at the time of hysterectomy performed for a benign indication: A retrospective review.

Accurate preoperative diagnoses are essential to allow surgeons to counsel patients preoperatively and manage intraoperative planning [ 1 , 2 ]. When presumably benign tissue is excised at surgery but the pathologic diagnosis is unexpectedly malignant, a new patient management plan must be created. Surgeons and oncologists must determine whether their patient has been properly staged or undergone necessary intraoperative procedures. For instance, patients with endometrial cancer require bilateral salpingo-oophorectomy, which may not have been done concomitantly with hysterectomy if the malignancy was unexpected [ 3 , 4 ]. Surgeons must also examine if any intra-operative procedures may have inadvertently upstaged the malignancy [ 5 ]. For example, intraoperative rupture of a benign ovarian cyst that is unexpectantly malignant upstages ovarian cancer [ 6 ] and intra-abdominal morcellation of malignant uterine tissue may facilitate growth and metastasis of cancer cells [ 7 ]. There is little research exploring the rate of unexpected malignancy at the time of hysterectomy performed for benign indications. Small retrospective studies have estimated the incidence between 0.25% [ 8 ] to 0.4% [ 9 ], although a larger American study of 6360 hysterectomies reported the rate of unexpected malignancy to be 2.7% [ 10 ]. To our knowledge, there is no literature on long-term assessment of hysterectomy procedures over multiple consecutive years among a Canadian patient population. The objective of this large-scale, multi-site retrospective review was to determine the proportion of patients undergoing hysterectomy for benign indications who had unexpected malignancy (UM) on postoperative pathology and to characterize the nature of UMs. We also aimed to determine factors associated with unexpected malignancy discovered post-hysterectomy.

A total of 3529 hysterectomies were performed during the study period. After exclusions (n = 750), 2779 hysterectomies completed for benign preoperative diagnoses were analyzed. On final pathology, 2728 (98.2%) cases were reported as benign and 51 cases had at least one unexpectedly malignancy. One patient had two malignancies, including appendiceal, making the total number of unexpected malignancies in our cohort 52. The incidence of UM was 1.8% (51/2779). The most common gynecologic malignancy locations were uterine (endometrial (27/52, 51.9%), sarcoma (13/52, 25%) and gestational trophoblastic neoplasia (1/52, 1.9%)), ovarian (6/52, 11.5%) and fallopian tube (4/52, 7.6%). There were no occult cervical cancers ( Fig 1 ). *51 patients had an unexpected malignancy with one patient having two malignancies (endometrial and appendiceal) (n = 52 total unexpected malignancies). Among patients with UM, the most common indications for surgery were fibroids (19/51, 37.2%) and abnormal uterine bleeding (16/51, 31.3%). Most patients with unexpected sarcoma had fibroids (12/13, 92.3%), while most with unexpected ovarian or fallopian tube cancer had pelvic mass as a preoperative diagnosis (6/10, 60%) ( Table 1 ). *Patients may have had more than one indication for surgery. GTN = gestational trophoblastic neoplasia. Compared to patients who had confirmed benign pathology, patients with UM were older (57.2 ± 11.4 years vs. 52.8 ± 12.5 years, p = .015) and had more previous laparotomies (2 (1.25, 2.0) vs. 1 (1.0, 1.0), p < .001). Higher BMI (29.7 ± 7.2 kg/m 2 vs. 28.0 ± 5.9 kg/m 2 , p = .049) and ASA class associated with UM (p < .028). ( Table 2 ) With respect to surgical factors, patients with UM had more adhesions (p = .001), transfusions (p = .020), and blood loss (p = .006) compared to those with benign pathology. Prevalence of UM was higher among patients undergoing abdominal hysterectomy (45.1% vs. 20.8%, p = .003) and lower among those undergoing vaginal hysterectomy (13.7% vs. 38.6%, p = .001). Surgeon training and case volume were not associated with UM ( Table 3 ). SD = standard deviation ASA = American Society of Anesthesiologists BMI = body mass index IQR = interquartile range *T-test **Fisher’s exact test ***Wilcox test. SD = standard deviation; IQR = interquartile range; MIGS = Minimally Invasive Gynecologic Surgery *T-test **Fisher’s exact test ***Wilcox test. The observed c-statistic for the multiple logistic regression model was .77 (95% bootstrap CI: .71 - .85), indicating good predictive accuracy of the model [ 13 ]. Patient characteristics most strongly associated with UM were age (OR 2.57, 95% CI 1.78–3.72, p < .001) and a preoperative diagnosis of pelvic mass (OR 2.76, 95% CI 1.11–6.20, p = .019). A preoperative diagnosis of prolapse, voiding dysfunction, or stress urinary incontinence was protective against UM (OR 0.12, 95% CI 0.05–0.3, p < .001) ( Table 4 ). a Wald test. BMI = body mass index.

In summary, incidence of unexpected malignancy among patients having hysterectomy for a benign indication was 1.8%, with the most common cancer type being endometrial cancer and sarcoma. Several perioperative variables are associated with an increased chance of unexpected malignancy at the time of hysterectomy.

We performed a multi-center retrospective study of all consecutive hysterectomy procedures at seven Ontario hospitals (4 academic, 3 community) between July 2016 and December 2019. Hysterectomies completed by gynecologic oncologists or those completed for a preoperative diagnosis of premalignant, malignant, invasive placentation, or uncertain diagnosis were excluded from the analysis. Data was extracted using a combination of health record coding (International Classification of Diseases, 10 th revision, ICD-10) and review of electronic medical records. Data from each site was entered into a central Research Electronic Capture (REDCap) registry. Data quality-assurance measures included real-time flagging of missing data and values outside pre-established ranges, random chart review of 10% of cases by an independent evaluator, and independent review and grading of each complication by two evaluators (research assistant and staff gynecologist). Research ethics board approval was obtained (20-0030-C). The exposure of interest was hysterectomies performed by gynecologists for benign indications. The primary outcome was the proportion of patients with unexpected malignancy at the time of hysterectomy performed for a benign indication. Unexpected malignancy was defined as a malignancy on final pathology report from a procedure without a malignant or premalignant (i.e. endometrial hyperplasia or cervical dysplasia) preoperative surgical indication. Incidence of unexpected malignancy was calculated by determining the number of patients with unexpected malignancy over the study period divided by the number of hysterectomy cases in our cohort. Gynecologic malignancies were characterized by type including cervical cancer, uterine cancer (endometrial and sarcoma), fallopian tube cancer, and ovarian cancer. Non-gynecologic malignancies (i.e. appendiceal cancer) were also recorded. We compared patients with unexpected malignancy to those without. We compared patient characteristics including age, body mass index (BMI), American Society of Anesthesiologist’s (ASA) class, previous abdominal or pelvic surgeries (laparotomy, laparoscopy or Caesarean section), and preoperative diagnoses. Surgical characteristics compared were route of hysterectomy (laparoscopic, vaginal, or abdominal), perioperative transfusion, estimated blood loss, presence of intraoperative endometriosis or adhesions, operative time, and uterine weight. We also recorded surgeon training and case volume. Surgeons were classified as either generalists (no additional training beyond residency) or fellowship-trained (Minimally Invasive Gynecologic Surgery (MIGS) or urogynecology/Pelvic Female Medicine and Reconstructive Surgery (FPMRS)). Fellowship training was considered additional training regardless of accreditation. Surgeon volume was defined as the mean number of hysterectomies performed over a 6-month period. Surgeons were considered low volume if they performed <6 cases per 6 months, average volume if 6 to 11 cases, and high volume if ≥ 12 cases. These cut-offs for surgical volume were informed by the literature [ 11 ]. Patient, surgical, and surgeon characteristics were compared between the benign and unexpected malignancy groups using standard bivariate methods (t-test, Wilcox rank sum test, chi-squared test, and Fisher’s exact test), as appropriate. Based on these findings, along with pragmatic and clinical considerations, we included several variables in the final multiple logistic regression model. It was anticipated the UM outcome was going to be infrequent, hence we were judicious in the covariates input into the model. Variables known to surgeons pre-operatively were selected as they would be most clinically meaningful with respect to preoperative counselling. We chose age, BMI, previous laparotomy, and selected preoperative diagnoses. We reported adjusted odds ratios (ORs), 95% confidence intervals (CIs), and Wald test p-values for the included variables. Patient age and BMI entered the model as continuous covariates and were scaled so the reported OR was for one standard deviation increase. ​The OR for previous laparotomy was expressed for a one-count increase in the number of previous laparotomies. We quantified the quality of model fit using the c-statistic with 95% confidence interval based on 1000 bootstrap samples. Alpha = .05 was adopted as the threshold for statistical significance. R version 3.6.2 software was used for the analysis [ 12 ].