Introduction

Ectopic pregnancy accounts for 1.5-2% of all pregnancies and is an important cause of morbidity and mortality in women of reproductive age. Early and effective treatment either with surgical or medical management is critical. Systemic methotrexate was first recognized as a medical treatment for unruptured ectopic pregnancy in 1982 by Tanaka et al, and it has since become widely accepted as a safe and effective alternative to surgery in the stable patient. Currently, there are 3 main treatment regimens for management of ectopic pregnancy with methotrexate: the multidose protocol, the single-dose protocol, and the 2-dose protocol. As the number and timing of methotrexate doses differ in these protocols, one may expect trends in serum human chorionic gonadotropin (hCG) and time to resolution to vary depending on protocol. By extension, hCG trends and time to resolution may predict ultimate treatment success. As such, this study aimed to evaluate the association between methotrexate protocol (single dose vs 2 doses), hCG trends, and time to resolution of ectopic pregnancy with the hypothesis that the 2-dose protocol would be associated with faster initial decline in serum hCG levels and a shorter time to resolution.

Materials and Methods

This prospective cohort study included clinical data collected at 3 academic medical centers from Aug. 1, 2007, through June 30, 2009: the University of Pennsylvania, the University of Miami, and the University of Southern California. The study was approved by the institutional review board at each of these institutions. Informed consent was obtained from all individual participants included in the study. Subjects were initially encountered both as emergency room and as emergency walk-in consultations, but all methotrexate was given on an obstetric outpatient basis. None of the subjects conceived using assisted reproductive technologies. Women who: (1) presented with first-trimester vaginal bleeding, pelvic pain, or both; (2) were diagnosed with ectopic pregnancy; and (3) underwent medical management with either the single-dose or 2-dose methotrexate protocols were included and followed up to assess treatment outcome. Diagnosis of ectopic pregnancy was made by ultrasound, abnormal serum hCG trend, and/or by the absence of products of conception after uterine evacuation according to American Congress of Obstetricians and Gynecologists guidelines. Women with nontubal ectopic pregnancies (ie, interstitial/cornual, cesarean delivery scar, cervical, intraabdominal, or ovarian) and heterotopic pregnancies were excluded from the analysis. Women initially treated via salpingostomy or who had initial serum hCG levels >10,000 mIU/mL were also excluded. Single-dose methotrexate was administered in accordance with the protocol originally described by Stovall et al in 1991. In brief, methotrexate is administered intramuscularly (IM) at a dose based on body surface area (50 mg/m ² ) on day 0. Serum hCG is then measured on posttreatment days 4 and 7. If at least a 15% decrease in hCG is observed between days 4-7, these women are then followed up with weekly hCG measurements until the result is negative. If the decline between days 4-7 is <15%, a second IM dose of methotrexate (50 mg/m ² ) is administered on day 7. Repeat hCG measurements are then obtained and if, during follow-up, hCG levels plateau or increase, methotrexate may be repeated.

The 2-dose methotrexate protocol was administered as described by Barnhart et al in 2007. According to this protocol, IM methotrexate of 50 mg/m ² is administered on days 0 and 4. As in the single-dose protocol, hCG is measured on days 4 and 7; if hCG does not decline at least 15% between days 4-7, a third dose of methotrexate is administered. Patients who receive a third dose of methotrexate return on day 11 for another hCG measurement. If the hCG level decreases by at least 15% between days 7-11, weekly hCG measurements are performed until a negative result is obtained. Otherwise, a fourth dose of methotrexate is administered, and another hCG level is obtained on day 14. If there is at least a 15% decrease between days 11-14, weekly hCG measurements are performed until a negative result is obtained. If at least a 15% decrease does not occur, the patient is referred for surgical management.

Baseline characteristics including clinical site, age, race, ethnicity, gravidity, parity, weight, body mass index, history of ectopic pregnancy, and history of spontaneous abortion were collected at initial presentation together with gestational age, initial hCG value, presence of pain and bleeding, ultrasound characteristics, method of diagnosis, and treatment outcome. Serum hCG values were collected from the day of the initial methotrexate dose (day 0, or T1), the first assessment after the initial dose (day 4 or T2), and the second assessment after the initial dose (day 7 or T3). The changes in hCG from day 0-4, day 4-7, and day 0-7 were obtained by calculating percent change between each of the 2 time points.

The primary outcome of interest was percent change in hCG from day 0-4, day 4-7, and day 0-7. Secondary outcomes included treatment success rates and time to successful resolution. Successful resolution was defined as achieving an hCG level of <5 mIU/mL, and lack of resolution was defined as needing definitive surgical management after treatment with methotrexate. Those who had lack of resolution or were lost to follow-up were considered censored for the event outcome of time to successful resolution. A sensitivity analysis was performed to assess the impact of loss to follow-up by defining successful resolution as a final hCG level of <100 mIU/mL, and lack of resolution as either needing definitive surgical management after treatment with methotrexate or having a final hCG level of >100 mIU/mL.

Propensity score modeling was utilized to address possible confounding by indication, ie, differential assignment of patients with better prognosis to the single-dose methotrexate protocol. A logistic model was developed to predict whether a patient was more likely to receive one intervention (ie, the 2-dose protocol) over the other (ie, the single-dose protocol). Forward stepwise variable selection was used to evaluate all available covariates (eg, clinical site, patient demographics, medical history) to develop the final propensity score model, which included initial hCG value, site, parity, method of ectopic pregnancy diagnosis, and ultrasound impression. Based on the model’s predicted probabilities of receiving the 2-dose protocol, each subject was assigned a propensity score, which was used in subsequent regression modeling.

Subjects were analyzed on the basis of treatment received, since intent to treat was not recorded. Differences in baseline characteristics between the 2 protocols were assessed by t tests (all continuous variables) and χ ² and proportion tests (all categorical variables). Associations between percent change in hCG outcomes, treatment protocol, and categorical covariates of interest were evaluated using linear regression models after adjustment for the propensity score. Similarly, Cox proportional hazards models were used to assess protocol and covariate differences in time to successful resolution. Models were also adjusted for any covariates that were still unbalanced across the protocols after propensity score adjustment, the only one of which was site of treatment. Statistical analyses were performed with SAS 9.2 (SAS Institute, Cary, NC) and STATA 12 (StataCorp LP, College Station, TX).

Materials and Methods

This prospective cohort study included clinical data collected at 3 academic medical centers from Aug. 1, 2007, through June 30, 2009: the University of Pennsylvania, the University of Miami, and the University of Southern California. The study was approved by the institutional review board at each of these institutions. Informed consent was obtained from all individual participants included in the study. Subjects were initially encountered both as emergency room and as emergency walk-in consultations, but all methotrexate was given on an obstetric outpatient basis. None of the subjects conceived using assisted reproductive technologies. Women who: (1) presented with first-trimester vaginal bleeding, pelvic pain, or both; (2) were diagnosed with ectopic pregnancy; and (3) underwent medical management with either the single-dose or 2-dose methotrexate protocols were included and followed up to assess treatment outcome. Diagnosis of ectopic pregnancy was made by ultrasound, abnormal serum hCG trend, and/or by the absence of products of conception after uterine evacuation according to American Congress of Obstetricians and Gynecologists guidelines. Women with nontubal ectopic pregnancies (ie, interstitial/cornual, cesarean delivery scar, cervical, intraabdominal, or ovarian) and heterotopic pregnancies were excluded from the analysis. Women initially treated via salpingostomy or who had initial serum hCG levels >10,000 mIU/mL were also excluded. Single-dose methotrexate was administered in accordance with the protocol originally described by Stovall et al in 1991. In brief, methotrexate is administered intramuscularly (IM) at a dose based on body surface area (50 mg/m ² ) on day 0. Serum hCG is then measured on posttreatment days 4 and 7. If at least a 15% decrease in hCG is observed between days 4-7, these women are then followed up with weekly hCG measurements until the result is negative. If the decline between days 4-7 is <15%, a second IM dose of methotrexate (50 mg/m ² ) is administered on day 7. Repeat hCG measurements are then obtained and if, during follow-up, hCG levels plateau or increase, methotrexate may be repeated.

The 2-dose methotrexate protocol was administered as described by Barnhart et al in 2007. According to this protocol, IM methotrexate of 50 mg/m ² is administered on days 0 and 4. As in the single-dose protocol, hCG is measured on days 4 and 7; if hCG does not decline at least 15% between days 4-7, a third dose of methotrexate is administered. Patients who receive a third dose of methotrexate return on day 11 for another hCG measurement. If the hCG level decreases by at least 15% between days 7-11, weekly hCG measurements are performed until a negative result is obtained. Otherwise, a fourth dose of methotrexate is administered, and another hCG level is obtained on day 14. If there is at least a 15% decrease between days 11-14, weekly hCG measurements are performed until a negative result is obtained. If at least a 15% decrease does not occur, the patient is referred for surgical management.

Baseline characteristics including clinical site, age, race, ethnicity, gravidity, parity, weight, body mass index, history of ectopic pregnancy, and history of spontaneous abortion were collected at initial presentation together with gestational age, initial hCG value, presence of pain and bleeding, ultrasound characteristics, method of diagnosis, and treatment outcome. Serum hCG values were collected from the day of the initial methotrexate dose (day 0, or T1), the first assessment after the initial dose (day 4 or T2), and the second assessment after the initial dose (day 7 or T3). The changes in hCG from day 0-4, day 4-7, and day 0-7 were obtained by calculating percent change between each of the 2 time points.

The primary outcome of interest was percent change in hCG from day 0-4, day 4-7, and day 0-7. Secondary outcomes included treatment success rates and time to successful resolution. Successful resolution was defined as achieving an hCG level of <5 mIU/mL, and lack of resolution was defined as needing definitive surgical management after treatment with methotrexate. Those who had lack of resolution or were lost to follow-up were considered censored for the event outcome of time to successful resolution. A sensitivity analysis was performed to assess the impact of loss to follow-up by defining successful resolution as a final hCG level of <100 mIU/mL, and lack of resolution as either needing definitive surgical management after treatment with methotrexate or having a final hCG level of >100 mIU/mL.

Propensity score modeling was utilized to address possible confounding by indication, ie, differential assignment of patients with better prognosis to the single-dose methotrexate protocol. A logistic model was developed to predict whether a patient was more likely to receive one intervention (ie, the 2-dose protocol) over the other (ie, the single-dose protocol). Forward stepwise variable selection was used to evaluate all available covariates (eg, clinical site, patient demographics, medical history) to develop the final propensity score model, which included initial hCG value, site, parity, method of ectopic pregnancy diagnosis, and ultrasound impression. Based on the model’s predicted probabilities of receiving the 2-dose protocol, each subject was assigned a propensity score, which was used in subsequent regression modeling.

Subjects were analyzed on the basis of treatment received, since intent to treat was not recorded. Differences in baseline characteristics between the 2 protocols were assessed by t tests (all continuous variables) and χ ² and proportion tests (all categorical variables). Associations between percent change in hCG outcomes, treatment protocol, and categorical covariates of interest were evaluated using linear regression models after adjustment for the propensity score. Similarly, Cox proportional hazards models were used to assess protocol and covariate differences in time to successful resolution. Models were also adjusted for any covariates that were still unbalanced across the protocols after propensity score adjustment, the only one of which was site of treatment. Statistical analyses were performed with SAS 9.2 (SAS Institute, Cary, NC) and STATA 12 (StataCorp LP, College Station, TX).