Original Research ajog.

org

GYNECOLOGY
Medical management of ectopic pregnancy with
single-dose and 2-dose methotrexate protocols: human
chorionic gonadotropin trends and patient outcomes
Michelle C. Mergenthal, MD; Suneeta Senapati, MD; Jarcy Zee, PhD; Lynne Allen-Taylor, PhD;
Paul G. Whittaker, DPhil; Peter Takacs, MD; Mary D. Sammel, ScD; Kurt T. Barnhart, MD

BACKGROUND: Ectopic pregnancy, although rare, is an important (30%) with the 2-dose protocol. Site, race, ethnicity, and reported pain
cause of female morbidity and mortality and early, effective treatment is level were associated with differential protocol allocation (P < .001,
critical. Systemic methotrexate has become widely accepted as a safe and P ¼ .011, P < .001, and P ¼ .035, respectively). Women had similar
effective alternative to surgery in the stable patient. As the number and initial human chorionic gonadotropin levels in either protocol but the
timing of methotrexate doses differ in the 3 main medical treatment mean rate of decline of human chorionic gonadotropin from day 0 (day
regimens, one might expect trends in serum human chorionic gonado- of administration of first dose of methotrexate) to day 7 was significantly
tropin and time to resolution to vary depending on protocol. Furthermore, more rapid in women who received the single-dose protocol compared
human chorionic gonadotropin trends and time to resolution may predict to those treated with the 2-dose protocol (mean change e31.3% vs
ultimate treatment success. e10.4%, P ¼ .037, adjusted for propensity score and site). The 2
OBJECTIVE: This study hypothesized that the 2-dose methotrexate protocols had no significant differences in success rate or time to
protocol would be associated with a faster initial decline in serum human resolution.
chorionic gonadotropin levels and a shorter time to resolution compared to CONCLUSION: In a racially and geographically diverse group of
the single-dose protocol. women, the single- and double-dose methotrexate protocols had com-
STUDY DESIGN: A prospective multicenter cohort study included parable outcomes. The more rapid human chorionic gonadotropin initial
clinical data from women who received medical management for ectopic decline in the single-dose group suggested these patients were probably
pregnancy. Rates of human chorionic gonadotropin change and successful at lower risk for ectopic rupture than those getting the 2-dose protocol. A
pregnancy resolution were assessed. Propensity score modeling addressed prospective randomized controlled design is needed to remove con-
confounding by indication, the potential for differential assignment of founding by indication.
patients with better prognosis to the single-dose methotrexate protocol.
RESULTS: In all, 162 ectopic pregnancies were in the final analysis; Key words: ectopic pregnancy, human chorionic gonadotropin,
114 (70%) were treated with the single-dose methotrexate and 48 methotrexate, protocol comparisons

Introduction regimens for management of ectopic medical centers from Aug. 1, 2007,
Ectopic pregnancy accounts for 1.5-2%1 pregnancy with methotrexate: the mul- through June 30, 2009: the University of
of all pregnancies and is an important tidose protocol, the single-dose proto- Pennsylvania, the University of Miami,
cause of morbidity and mortality in col,5 and the 2-dose protocol.6 As the and the University of Southern Califor-
women of reproductive age. Early and number and timing of methotrexate nia. The study was approved by the
effective treatment either with surgical doses differ in these protocols, one may institutional review board at each of
or medical management is critical. Sys- expect trends in serum human chorionic these institutions. Informed consent
temic methotrexate was first recognized gonadotropin (hCG) and time to reso- was obtained from all individual partic-
as a medical treatment for unruptured lution to vary depending on protocol. By ipants included in the study. Subjects
ectopic pregnancy in 1982 by Tanaka extension, hCG trends and time to res- were initially encountered both as
et al,2 and it has since become widely olution may predict ultimate treatment emergency room and as emergency
accepted as a safe and effective alterna- success. As such, this study aimed to walk-in consultations, but all metho-
tive to surgery in the stable patient.3,4 evaluate the association between meth- trexate was given on an obstetric outpa-
Currently, there are 3 main treatment otrexate protocol (single dose vs 2 tient basis. None of the subjects
doses), hCG trends, and time to resolu- conceived using assisted reproductive
tion of ectopic pregnancy with the technologies. Women who: (1) presented
Cite this article as: Mergenthal MC, Senapati S, Zee J, hypothesis that the 2-dose protocol with first-trimester vaginal bleeding,
et al. Medical management of ectopic pregnancy with would be associated with faster initial pelvic pain, or both; (2) were diagnosed
single-dose and 2-dose methotrexate protocols: human decline in serum hCG levels and a with ectopic pregnancy; and (3) under-
chorionic gonadotropin trends and patient outcomes. Am
shorter time to resolution. went medical management with either
J Obstet Gynecol 2016;215:590.e1-5.
the single-dose or 2-dose methotrexate
0002-9378/$36.00 Materials and Methods protocols were included and followed up
ª 2016 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.ajog.2016.06.040 This prospective cohort study included to assess treatment outcome. Diagnosis
clinical data collected at 3 academic of ectopic pregnancy was made by

590.e1 American Journal of Obstetrics & Gynecology NOVEMBER 2016

ajog.org GYNECOLOGY Original Research

ultrasound, abnormal serum hCG trend, Baseline characteristics including parity, method of ectopic pregnancy
and/or by the absence of products of clinical site, age, race, ethnicity, diagnosis, and ultrasound impression.
conception after uterine evacuation gravidity, parity, weight, body mass in- Based on the model’s predicted proba-
according to American Congress of dex, history of ectopic pregnancy, and bilities of receiving the 2-dose protocol,
Obstetricians and Gynecologists guide- history of spontaneous abortion were each subject was assigned a propensity
lines.4 Women with nontubal ectopic collected at initial presentation together score, which was used in subsequent
pregnancies (ie, interstitial/cornual, with gestational age, initial hCG value, regression modeling.6,7
cesarean delivery scar, cervical, intra- presence of pain and bleeding, ultra- Subjects were analyzed on the basis of
abdominal, or ovarian) and heterotopic sound characteristics, method of diag- treatment received, since intent to treat
pregnancies were excluded from the nosis, and treatment outcome. Serum was not recorded. Differences in baseline
analysis. Women initially treated via sal- hCG values were collected from the day characteristics between the 2 protocols
pingostomy or who had initial serum of the initial methotrexate dose (day 0, or were assessed by t tests (all continuous
hCG levels >10,000 mIU/mL were also T1), the first assessment after the initial variables) and c2 and proportion tests
excluded. Single-dose methotrexate was dose (day 4 or T2), and the second (all categorical variables). Associations
administered in accordance with the assessment after the initial dose (day 7 or between percent change in hCG out-
protocol originally described by Stovall T3). The changes in hCG from day 0-4, comes, treatment protocol, and cate-
et al5 in 1991. In brief, methotrexate is day 4-7, and day 0-7 were obtained by gorical covariates of interest were
administered intramuscularly (IM) at calculating percent change between evaluated using linear regression models
a dose based on body surface area each of the 2 time points. after adjustment for the propensity
(50 mg/m2) on day 0. Serum hCG is then The primary outcome of interest was score. Similarly, Cox proportional haz-
measured on posttreatment days 4 and 7. percent change in hCG from day 0-4, ards models were used to assess protocol
If at least a 15% decrease in hCG is day 4-7, and day 0-7. Secondary and covariate differences in time to
observed between days 4-7, these women outcomes included treatment success successful resolution. Models were also
are then followed up with weekly hCG rates and time to successful resolution. adjusted for any covariates that were still
measurements until the result is negative. Successful resolution was defined as unbalanced across the protocols after
If the decline between days 4-7 is <15%, achieving an hCG level of <5 mIU/mL, propensity score adjustment, the only
a second IM dose of methotrexate and lack of resolution was defined as one of which was site of treatment. Sta-
(50 mg/m2) is administered on day 7. needing definitive surgical management tistical analyses were performed with
Repeat hCG measurements are then after treatment with methotrexate. SAS 9.2 (SAS Institute, Cary, NC) and
obtained and if, during follow-up, hCG Those who had lack of resolution or STATA 12 (StataCorp LP, College
levels plateau or increase, methotrexate were lost to follow-up were considered Station, TX).
may be repeated.5 censored for the event outcome of time
The 2-dose methotrexate protocol was to successful resolution. A sensitivity Results
administered as described by Barnhart analysis was performed to assess the In all, 162 ectopic pregnancies were
et al6 in 2007. According to this protocol, impact of loss to follow-up by defining included in the final analysis; 114
IM methotrexate of 50 mg/m2 is successful resolution as a final hCG (70.4%) were treated with the single-
administered on days 0 and 4. As in the level of <100 mIU/mL, and lack of dose methotrexate protocol and 48
single-dose protocol, hCG is measured resolution as either needing definitive (29.6%) with the 2-dose protocol. At day
on days 4 and 7; if hCG does not decline surgical management after treatment 7, data were available on 106 patients
at least 15% between days 4-7, a third with methotrexate or having a final treated with the single-dose metho-
dose of methotrexate is administered. hCG level of >100 mIU/mL. trexate protocol and 42 with the 2-dose
Patients who receive a third dose of Propensity score modeling was uti- protocol (retention 93% and 88%,
methotrexate return on day 11 for lized to address possible confounding by respectively). Baseline characteristics of
another hCG measurement. If the hCG indication, ie, differential assignment of the 2 groups are described in Table 1.
level decreases by at least 15% between patients with better prognosis to the The use of single-dose vs 2-dose protocol
days 7-11, weekly hCG measurements single-dose methotrexate protocol. A was significantly associated with site,
are performed until a negative result is logistic model was developed to predict race, ethnicity, and reported pain level (P
obtained. Otherwise, a fourth dose whether a patient was more likely to < .001, P ¼ .011, P < .001, and P ¼ .035,
of methotrexate is administered, and receive one intervention (ie, the 2-dose respectively). Patients treated at the
another hCG level is obtained on day 14. protocol) over the other (ie, the single- University of Pennsylvania were signifi-
If there is at least a 15% decrease between dose protocol). Forward stepwise vari- cantly more likely to receive the 2-dose
days 11-14, weekly hCG measurements able selection was used to evaluate all vs single-dose protocol (79.6% vs
are performed until a negative result is available covariates (eg, clinical site, pa- 20.4%, P < .001), whereas patients
obtained. If at least a 15% decrease does tient demographics, medical history) to treated at the University of Miami and
not occur, the patient is referred for develop the final propensity score model, the University of Southern California
surgical management.6 which included initial hCG value, site, were more likely to receive the

NOVEMBER 2016 American Journal of Obstetrics & Gynecology 590.e2

Original Research GYNECOLOGY ajog.org

single-dose vs 2-dose protocol (97.5% vs

TABLE 1
2.5%, P <.001 and 78.1% vs 21.9%, P ¼
Baseline characteristics of study population across 3 sites
.002, respectively). Caucasian patients
(83.6% vs 16.4%, P < .001), those who Single dose Two-dose
reported race as “other” (69.2% vs MTX, n ¼ 114 MTX, n ¼ 48
30.8%, P ¼ .050), and Hispanic patients Age, y, mean ( SD) 30.6 (6.3) 30.1 (6.9)
(87% vs 13%, P < .001) were signifi- a
Site (%)
cantly more likely to receive the single-
dose vs 2-dose protocol. Patients who University of Miamia 79 (97.5) 2 (2.5)
a
reported no pain, mild pain, or moder- University of Pennsylvania 10 (20.4) 39 (79.6)
ate pain at presentation were more likely University of Southern California b
25 (78.1) 7 (21.9)
to receive the single-dose protocol over c
Race (%)
the 2-dose protocol (P ¼ .035, P < .001,
and P ¼ .001, respectively). There were African American 45 (60.0) 30 (40.0)
c
no significant differences with respect Other 18 (69.2) 8 (30.8)
to age, gravidity, parity, body mass Caucasian a
51 (83.6) 10 (16.4)
index, history of ectopic pregnancy, a
Ethnicity (%)
history of spontaneous abortion, gesta-
tional age, or severity of vaginal bleeding Hispanica 60 (87.0) 9 (13.0)
at presentation between the 2 protocol Non-Hispanic 54 (58.1) 39 (41.9)
groups. Gravidity, mean ( SD) 2.9 (1.9) 2.9 (1.6)
Women had similar initial hCG
Parity, mean ( SD) 0.9 (1.0) 1.0 (1.1)
levels in either protocol (Table 1). When
we examined hCG trends by treat- BMI, mean ( SD) 30.1 (6.4) 29.1 (5.5)
ment protocol using linear regression Prior ectopic (%) 22 (78.6) 6 (21.4)
modeling and adjusting for propensity Prior SAB (%) 38 (71.7) 15 (28.3)
score and site (Table 2), we found that
the mean decline of hCG from day Gestational age, wk, mean ( SD) 6.0 (2.8) 6.3 (2.1)
0 (day of administration of first dose of Initial hCG, mIU/mL, mean ( SD) 1684 (1922) 1839 (2112)
methotrexate) to day 4 was significantly Bleeding (%)
greater in women who were treated with
None 25 (21.9) 14 (29.2)
the single-dose protocol compared to
those treated with the 2-dose protocol Mild 73 (64.0) 30 (62.5)
(mean change e10.8% vs þ5.14%, P ¼ Moderate/severe 16 (14.0) 4 (8.3)
.031). Of note, the average serum hCG c
Pain (%)
level decreased from day 0-4 in the group
Nonec 29 (65.9) 15 (34.1)
that received the single-dose protocol, a
while it increased in the group that Mild 43 (82.7) 9 (17.3)
b
received the 2-dose protocol during Moderate 36 (73.5) 13 (26.5)
the same time period. The hCG levels Severe 3 (37.5) 5 (62.5)
also declined faster in the single-dose Data are mean (SD) or n (%).
compared to the 2-dose group from BMI, body mass index; hCG, human chorionic gonadotropin; MTX, methotrexate; SAB, spontaneous abortion.
day 0-7 (mean change e31.3% vs a
P <.001; b P <.01; c P <.05 significant statistical difference between protocols (Student t test or c2 tests).
e10.4%, P ¼ .037). There was no sig- Mergenthal et al. One- and 2-dose methotrexate protocols for ectopic pregnancy. Am J Obstet Gynecol 2016.
nificant difference in percent change
from day 4-7 between single-dose and
2-dose protocols after adjusting for
propensity score and site (mean change (Table 3). Time to resolution of hCG A sensitivity analysis was performed
e31.8% vs e10.6%, P ¼ .110). from the serum for women in each to assess the impact of loss to follow-up.
Rates of treatment success were com- protocol is presented in the Figure. In this analysis, resolution was defined as
parable between the 2 groups (83% for There were also no distinguishing factors a final hCG level of <100 mIU/mL, and
single-dose vs 79% for 2-dose, P ¼ .53). between successful and failed therapy lack of resolution as either needing
There was no significant difference among the women of each treatment definitive surgical management after
between the 2 groups in time to resolu- group. The number of women needing treatment with methotrexate or having
tion (hazard ratio, 0.441; 95% confi- urgent intervention was not different a final hCG level of >100 mIU/mL. As
dence interval, 0.139e1.401; P ¼ .165) between the 2 groups. in the original analysis, there was no

590.e3 American Journal of Obstetrics & Gynecology NOVEMBER 2016

ajog.org GYNECOLOGY Original Research

minimize differences in outcome be-

TABLE 2
tween groups due to confounding.
Human chorionic gonadotropin trends by methotrexate protocol
Adjustment with the propensity score
Single-dose Two-dose demonstrated that the magnitude of
methotrexate methotrexate differences in initial hCG trends, before
Unadjusted Adjusted
Geometric mean (95% confidence interval) P valuea P valueb the second dose was given, was even
greater than without adjustment
% D T1eT2 e10.8 (e20.2 to e1.3) 5.1 (e15.4 to 25.7) .109 .031
(Table 2). This suggested that differences
% D T2eT3 e31.8 (e37.4 to e26.1) e10.6 (e50.7 to 29.5) .123 .110 in hCG trends prior to treatment was
% D T1eT3 e31.3 (e41.8 to e20.7) e10.4 (e32.0 to 11.2) .055 .037 not due to differences in population
% D, percent change between time points; T1, day 0; T2, day 4; T3, day 7. characteristics (which are included in the
a
Comparison value from linear regression modeling; b Adjusted for propensity score and site. propensity score) but rather confound-
Mergenthal et al. One- and 2-dose methotrexate protocols for ectopic pregnancy. Am J Obstet Gynecol 2016. ing by indication (choice or protocol
by the clinician). A propensity score
adjustment cannot control for the like-
significant difference between the 2 prognosis. Evidence for this is the dra- lihood that clinicians who are aware of
groups in time to resolution (hazard matic difference (between groups) in the entire clinical picture choose a
ratio, 0.531; 95% confidence interval, hCG values between day 0 (day of 2-dose protocol for women at higher risk
0.232e1.215; P ¼ .134) (Table 3). methotrexate administration) and day 4. based on clinical signs and symptoms,
In women who received the 2-dose despite a similar baseline hCG level. For
Comment protocol, the day-4 hCG value (ob- example, the trend in hCG levels prior
We had hypothesized that women tained prior to administration of the to treatment of methotrexate was not
who received the 2-dose protocol second dose of methotrexate) was actu- available in our analysis, but could affect
would experience a greater hCG ally increased, which suggested that these both the choice of single- vs 2-dose
decline and a faster time to ultimate women had a more aggressive ectopic methotrexate and the trajectory of
successful resolution when compared pregnancy. Ultimately, there was no hCG after treatment. Alternatively it is
to those receiving the single-dose pro- difference in success rates or time to conceivable that clinicians noted the
tocol (who receive only 1 dose of resolution between the 2 groups, sug- dramatic decline in hCG between days
methotrexate in the first 7 days), gesting that the 2-dose protocol may be 0-4 and thus withheld the second dose
because patients in the 2-dose protocol attenuating the differences in prognosis of methotrexate (even though they
receive more methotrexate, sooner. and subsequent risk of failure. may have intended to use the 2-dose
However, we did not observe this effect. To address the issue of confounding protocol).
Instead, counterintuitively, we found by indication, we developed a propensity One of the major strengths of our
that hCG levels declined more rapidly score model to predict whether a patient study was that data were collected at 3
from day 0-4 and from day 0-7 in those was more likely to receive one metho- large institutions located in different
receiving the single-dose protocol. This trexate protocol over the other, and then regions of the country, and our subjects
unexpected finding was noted despite applied that model to subsequent ana- represented a racially and ethnically
similar initial hCG levels, and no lyses. The incorporation of propensity diverse group of women. We would
difference in success rate or time to score modeling to study the association expect this to increase the generaliz-
successful resolution. between methotrexate protocol and ability of our results. Our study also
These findings suggest the occurrence various outcomes is unique in the could have been impacted by informa-
of confounding by indication in that ectopic pregnancy literature. It would be tion bias; however, the risk of this was
the single-dose protocol may have anticipated that controlling for pro- minimized through use of a clinical
been reserved for patients with better pensity in a multivariable analysis would database in which data were prospec-
tively collected. Another potential limi-
tation is that not all women were
TABLE 3 followed up until hCG was completely
Time to resolution, adjusted for propensity score cleared from the serum, although there
were no reports of failed treatment with
N N (eventsa) HR 95% CI P very low hCG values. When assessing
Original analysis 162 68 0.441 0.139e1.401 .165 time to resolution, this study treated
Sensitivity analysis 162 133 0.531 0.232e1.215 .134 those who were lost to follow-up as
CI, confidence interval; HR, hazard ratio.
censored. However, the sensitivity anal-
a
Successes.
ysis suggests that these findings are
Mergenthal et al. One- and 2-dose methotrexate protocols for ectopic pregnancy. Am J Obstet Gynecol 2016. robust to variation in final hCG level
up to 100 mIU/mL.

NOVEMBER 2016 American Journal of Obstetrics & Gynecology 590.e4

Original Research GYNECOLOGY ajog.org

United States, 1991-1999. MMWR Surveill

FIGURE Summ 2003;52:1-8.
Cox model survival estimates of time to resolution 2. Tanaka T, Hayashi H, Kutsuzawa T,
Fujimoto S, Ichinoe K. Treatment of interstitial
ectopic pregnancy with methotrexate: report of
a successful case. Fertil Steril 1982;37:851-5.
3. Barnhart KT. Ectopic pregnancy. N Engl J
Med 2009;361:379-87.
4. American College of Obstetricians and Gy-
necologists. Medical management of ectopic
pregnancy. ACOG Practice bulletin no. 94.
Obstet Gynecol 2008;111:1479-85.
5. Stovall TG, Ling FW, Gray LA. Single-dose
methotrexate for treatment of ectopic preg-
nancy. Obstet Gynecol 1991;77:754-7.
6. Barnhart K, Hummel AC, Sammel MD,
Menon S, Jain J, Chakhtoura N. Use of “2-dose”
regimen of methotrexate to treat ectopic preg-
nancy. Fertil Steril 2007;87:250-6.
7. Hamed HO, Ahmed SR, Alghasham AA.
Comparison of double- and single-dose meth-
otrexate protocols for treatment of ectopic
pregnancy. Int J Gynaecol Obstet 2012;116:
67-71.
8. Song T, Kim MK, Kim ML, Jung YW, Yun BS,
Seong SJ. Single-dose versus two-dose
administration of methotrexate for the treat-
ment of ectopic pregnancy: a randomized
controlled trial. Hum Reprod 2016;31:332-8.
Broken line indicates single-dose methotrexate protocol, solid line indicates 2-dose protocol. 9. Gungorduk K, Asicioglu O, Yildirim G,
Mergenthal et al. One- and 2-dose methotrexate protocols for ectopic pregnancy. Am J Obstet Gynecol 2016. Gungorduk OC, Besimoglu B, Ark C. Compari-
son of single-dose and two-dose methotrexate
protocols for the treatment of unruptured
Many of the limitations of the current the 2-dose protocol.7,8 Overall our ectopic pregnancy. J Obstet Gynaecol 2011;31:
study could be successfully addressed finding of similar success rates and 330-4.
with a randomized controlled trial time to resolution for the 2 protocols,
comparing hCG trends in single-dose despite confounding by indication,
and 2-dose methotrexate protocols as is consistent with the results of these Author and article information
From the Departments of Obstetrics and Gynecology (Drs
a possible mechanism for improved randomized trials7-9 and supports
Mergenthal, Senapati, Whittaker, Sammel, and Barnhart)
success. Small randomized trials exist the use of the 2-dose protocol as and Biostatistics and Epidemiology (Dr Zee, Taylor,
that looked at the efficacy between primary medical management of ectopic Whittaker, Sammel, and Barnhart), Perelman School of
single-dose and 2-dose protocols.7-9 pregnancy. n Medicine, University of Pennsylvania, Philadelphia, PA;
Individually all report no differences in and Department of Obstetrics and Gynecology, Eastern
Acknowledgment Virginia Medical School, Norfolk, VA (Dr Takacs).
treatment success based on protocol, but
Received May 16, 2016; revised June 18, 2016;
all report a trend toward better outcome The authors acknowledge contributions by
accepted June 20, 2016.
with the 2-dose protocol. When com- Karine Chung, MD, MSCE (Department of Ob-
This study was funded by National Institutes of
stetrics and Gynecology, University of Southern
bined, the superiority of the 2-dose Health grants K24HD060687 (K.T.B., M.D.S.) and
California), to data collection and interpretation.
protocol approaches statistical signifi- R01HD076279 (K.T.B., M.D.S.). The sponsor had no role
in data collection, analysis, interpretation, report writing,
cance (83.5% vs 88.5% success: relative
or decision to publish.
risk, 0.94 confidence interval; References The authors report no conflict of interest.
0.86e1.03).7 Two of these studies also 1. Chang J, Elam-Evans LD, Berg CJ, et al. Corresponding author: Kurt T. Barnhart, MD.
find a reduced time to resolution with Pregnancy-related mortality surveillancee [email protected]

590.e5 American Journal of Obstetrics & Gynecology NOVEMBER 2016