A validation of the most commonly used protocol to predict the success of single-dose methotrexate in the treatment of ectopic pregnancy

E. Kirk1,5, G. Condous1,2, B. Van Calster3, Z. Haider1, S. Van Huffel3, D. Timmerman4 and T. Bourne1

1 Department of Obstetrics & Gynaecology, Early Pregnancy, Gynaecological Ultrasound and MAS Unit, St George’s, University of London, London, UK 2 Early Pregnancy Unit, Nepean Hospital, Western Clinical School, Nepean Campus, University of Sydney, Sydney, Australia 3 Department of Electrical Engineering (ESAT-SISTA), K.U.Leuven and 4 Department of Obstetrics and Gynaecology, University Hospital Gasthuisberg, K.U.Leuven, Leuven, Belgium
5 To whom correspondence should be addressed at: Early Pregnancy, Gynaecological Ultrasound and MAS Unit, St George’s, University of London, Cranmer Terrace, London SW17 0RE, UK. E-mail: ejkirk@hotmail.co.uk
'//-->


Abstract
TopAbstractIntroductionMethodsResultsDiscussionReferences BACKGROUND: Currently, the likely success of single-dose methotrexate (MTX) (50 mg/m2) for the treatment of ectopic pregnancy is indicated by a >15% decrease in hCG from days 4–7 after administration. The aim of this study was to assess this protocol and to develop new rules that could be used to predict the outcome at an earlier stage. METHODS: Data were collected prospectively. Women receiving MTX for an ectopic pregnancy had serum hCG and progesterone levels checked on days 1, 3, 4, 5 and 7. Other factors including age, gestational age, previous obstetric history and ultrasound findings were recorded. The women were followed up until the outcome of medical management was known. Univariate analysis was performed to determine the benefit of the ‘15% day 4–7 rule’, as well as to develop new rules, which potentially could be used to predict the likelihood of success before 7 days. Historical and ultrasound variables were also analysed to identify the significant variables associated with successful conservative management. RESULTS: The success rate of single-dose MTX was 68.1% (47/69). A second dose was required in 18.8% (13/69) of cases, and it was successful in 84.6% (11/13). The 15% day 4–7 rule correctly predicted the outcome in 90.3% of cases [sensitivity 93.0%, specificity 84.2%, positive predictive value (PPV) 93.0% and negative predictive value (NPV) 84.2%, Fisher exact test P-value < 0.0001]. New rules were developed based on the percentage change day 4–5 and logistic regression models incorporating day 5 hCG levels and ultrasound findings. These new rules did not outperform the current 15% day 4–7 rule. CONCLUSIONS: We have confirmed that a 15% decrease in serum hCG between day 4 and day 7 is a very good indicator of the likely success of MTX. The development of new rules did not significantly improve our ability to predict a successful outcome at an earlier stage.
Key words: methotrexate/hCG/ectopic pregnancy/predictive value

Introduction
TopAbstractIntroductionMethodsResultsDiscussionReferences The anti-folate cytotoxic drug methotrexate (MTX) has been used for the management of ectopic pregnancies since the 1980s. It can be given orally, locally (either at the time of laparoscopy or under ultrasound guidance) or, more commonly, systemically. Systemic MTX can be administered according to either a multiple-dose or a single-dose regimen. In the UK, the single-dose regimen is favoured. It is as effective as multidose therapy and requires fewer MTX doses (Lipscomb et al., 2005). The highest current reported success rates for single-dose MTX are ~91% (Lipscomb et al., 2004).
The first single-dose MTX protocol was developed in 1991 (Stovall et al., 1991). The protocol involved giving women with an ectopic pregnancy: an i.m. injection of MTX at a dose of 50 mg/m2. Serum hCG levels were checked on day 1 (day of MTX) and days 2, 4 and 7 post treatment. The protocol stipulated that if there was a <15% href="http://humrep.oxfordjournals.org/cgi/content/full/22/3/858#B11">). The day 2 hCG level was however eliminated. All further published clinical studies on the use of single-dose MTX appear to use this protocol. However, no one to date has published data relating to the behaviour of hCG following MTX therapy to validate Stovall’s original rule. They have all simply applied the previous protocol. The aim of this study was to evaluate the 15% day 4–7 rule and to confirm that it correctly identifies those who will have successful management after single-dose MTX. New rules for the prediction of outcome from single-dose MTX before day 7 were also generated as a means of comparison. Other biochemical markers as well as historical factors and scan findings were also evaluated. The concept of predicting the likely success of MTX before 7 days is appealing because earlier administration of a second dose may result in fewer complications. It may also have the advantage of reducing follow-up time and the number of follow-up visits in those who are thought likely to be successful.

Methods
TopAbstractIntroductionMethodsResultsDiscussionReferences This was a prospective non-interventional observational study. All women treated with single-dose MTX in the Early Pregnancy Unit at St George’s Hospital were included. The reasons for MTX administration included tubal ectopic pregnancies, interstitial ectopic pregnancies, persisting pregnancies of unknown locations (PULs) and persistent trophoblastic tissue (PTT) after salpingotomy. A PUL is a clinical entity that can be defined by a positive pregnancy test with no signs of an intrauterine or extrauterine pregnancy on transvaginal ultrasonography (TVS). Persisting PULs are those PULs when the serum hCG level fails to decline, when there is no evidence of trophoblastic disease and the location of the pregnancy cannot be identified by TVS, laparoscopy or curettage (Condous et al., 2004). Although they behave biochemically like ectopic pregnancies, the final diagnosis is not known. PTT after a laparoscopic salpingotomy was defined as a failure of the hCG levels to decrease after surgery.
All women with an ectopic pregnancy who consented to the study were treated with single-dose MTX in the Early Pregnancy Unit at St George’s Hospital. Women initially received a single i.m. injection of MTX at a dose of 50 mg/m2 according to the protocol advocated by Stovall (Stovall and Ling, 1993). If the hCG levels decreased by 15% between days 4 and 7, then the levels were measured on a weekly basis until they reached 15 IU/l. If the hCG decreased by <15% between days 4 and 7, a second dose of MTX was given, provided the clinical situation allowed and the woman was in agreement. Serum hCG and progesterone levels were also checked on days 1, 3, 4, 5 and 7 in accordance with the study’s protocol. Serum hCG (World Health Organization, Third International Reference 75/537) and progesterone (Roche Elecsys 2010 Progesterone II test, Roche Diagnostics, Lewes, UK) levels were quantified using automated electrochemiluminescence immunoassays. All women who were selected for MTX therapy were followed up until the outcome of medical management was known. Treatment success was defined as a reduction in the hCG level to 15 IU/l without surgical intervention.
General clinical inclusion criteria for women receiving MTX were haemodynamic stability, no abdominal pain and normal kidney and liver function. Ultrasonographic inclusion criteria were no fetal cardiac activity and no haemoperitoneum on TVS.
Exclusion criteria for medical management included haemodynamic instability, an acute abdomen, haemoperitoneum on TVS, positive fetal cardiac activity, liver or renal impairment and probable poor compliance with conservative management. A threshold serum hCG level of 5000 IU/l was used for women with tubal ectopic pregnancies. For interstitial ectopic pregnancies, persisting PULs and PTT after salpingotomy, no upper limit of hCG was used.
Age, gestation, size of the ectopic pregnancy and time from presentation to diagnosis or administration of MTX were recorded for all women. All women were managed as outpatients, and if at any point the woman developed pain, she was reassessed to see whether a laparoscopy should be performed. All women were followed up in clinic 3 months after presentation.
Mann–Whitney tests were used to test for differences between the successful and the unsuccessful outcomes with respect to continuous variables whilst Fisher exact tests were used for the categorical variables. A P-value <0.05 href="http://humrep.oxfordjournals.org/cgi/content/full/22/3/858#B5">). Receiver operator characteristic (ROC) curves were constructed, as they can be interpreted as the probability of a variable or model to correctly distinguish between a successful and an unsuccessful case (Hanley and McNeil, 1982). The statistical analyses were conducted with SAS 9.1 (Statistical Analysis System Version 9.1, Cary, NC, USA).

Results
TopAbstractIntroductionMethodsResultsDiscussionReferences During the study period, April 2004 to October 2005, 69 women received MTX. There were 44 tubal ectopic pregnancies, representing 25.9% (44/170) of all tubal ectopic pregnancies treated in the time period, 3 interstitial ectopic pregnancies, 18 persisting PULs and 4 PTTs.
The success rate for a single dose of MTX was 68.1% (47/69). Of these 47 women, 42 (89%) had a >15% decrease in hCG between days 4–7. Five women (11%) did not have a documented 15% decrease in hCG day 4–7 and did not have a second dose of MTX but went on to have a successful outcome. Of these, two women had a suboptimal decrease in hCG (8 and 13%) but refused a second dose of MTX. The other three women failed to have blood samples taken on day 4, although they all had a decrease in hCG between days 5 and 7. Thirteen women (18.8%) required a second dose, and it was successful in 11 (84.6%). The overall success rate for systemic MTX was 84.1% (58/69). All 11 women needing surgery had tubal ectopic pregnancies and underwent salpingectomies—90.9% (10/11) were performed laparoscopically. The indications for surgery were worsening abdominal pain and haemoperitoneum on TVS in seven women, positive fetal cardiac activity on repeat TVS in one woman and individual choice to decline a second dose of MTX in three women. In the seven women with haemoperitoneum on TVS, only three were found to have tubal rupture at the time of surgery. All women made an uneventful recovery.
Day 4–7 ruleStovall’s rule correctly identified 90.3% (56/62) of cases that would go on to have a successful or unsuccessful outcome with MTX—sensitivity 93.0%, specificity 84.2%, positive predictive value (PPV) 93.0% and negative predictive value (NPV) 84.2%. There was no significant difference in the performance of this rule for either visualized ectopic pregnancies or PULs. The rule correctly identified 93.3% of tubal ectopic pregnancies that would go on to have successful or unsuccessful outcomes and 92.3% of PULs.
New rulesTable I summarizes descriptive statistics for the continuous variables recorded for both those with successful and those with unsuccessful outcomes after a single dose of MTX. Table II summarizes the Mann–Whitney test results for differences in these variables between those with successful and unsuccessful outcomes. Other variables that may help separate those with successful response to single-dose MTX were examined. Although, progesterone levels on days 1, 3, 4, 5 and 7 were significantly different between those with successful and unsuccessful outcomes, changes in the levels between the days were not (Table II). Parity, previous history of a miscarriage, termination of pregnancy or ectopic pregnancy and scan findings were not significant (Table III). The hCG levels were significantly different between the two outcome groups. The TVS findings, when the ectopic pregnancy was diagnosed, was not statistically significant, but the P-value was 0.0530.

View this table:[in this window][in a new window]
Table I. Descriptive statistics for the continuous variables in those receiving single-dose methotrexate (MTX)

View this table:[in this window][in a new window]
Table II. Mann–Whitney (MW) test results for differences in variables between those with successful and unsuccessful single-dose methotrexate management

View this table:[in this window][in a new window]
Table III. Significance of other variables in correctly separating those with successful and unsuccessful methotrexate treatment
When looking at the differences in hCG levels at the different times, days 1, 3, 4, 5 and 7, and over the 48h before MTX administration, between the successful and unsuccessful outcome groups, it can be seen that as time increases the P-value decreases (Table II). The percentage change in hCG from one day to another was then examined in both outcome groups and then levels of significance obtained (Table IV). It can be seen that all P-values involving day 7 were <0.0001. The best significance level, reached without waiting for the hCG level on day 7, was between days 4 and 5 (P-value < 0.0001). Because of this, a rule was developed using a threshold of a 3% decrease in serum hCG between days 4 and 5 to separate successful and unsuccessful cases. This rule correctly separated 83.3% (50/60) of cases, sensitivity 81.0%, specificity 88.9%, PPV 94.4% and NPV 66.7%. Using a range of thresholds from a 1–4% decrease gave sensitivities and specificities of at least 70%.

View this table:[in this window][in a new window]
Table IV. Significance of percentage change in hCG level from one day to another between the two outcome groups
Indices were also combined to develop models to predict likely success. Logistic regression analysis suggested that using only hCG indices collected before day 7 was not sufficient for improved performance. A logistic regression model was created including the ultrasound appearance of the ectopic pregnancy as a variable, together with the hCG level at day 5 (log transformed because of skewness) and the percentage change from day 4 to 5 (n = 54). The use of a threshold level of 0.66 for this model gave an accuracy of 87%, a sensitivity of 87%, a specificity of 88%, a PPV of 94% and an NPV of 74%.
ROC curves were created, because the area under such a curve can be interpreted as the probability that a pair (successful and unsuccessful) will be correctly separated based on a given criterion (Figure 1). This allowed comparisons to be made between the original day 4–7 rule and these new rules. The area under the curves (AUCs) were 0.845 using just the day 4–5 rule (n = 60) and 0.929 using the logistic regression model incorporating the ultrasound appearance of the ectopic (n = 54). The AUC for the original day 4–7 rule was 0.917 (n = 62).

View larger version (16K):[in this window][in a new window][Download PowerPoint slide]
Figure 1. A receiver operator characteristic (ROC) curve showing the probability that the original rule and the developed rules will correctly separate those with successful single-dose methotrexate treatment from those with unsuccessful treatment.
Further rules were developed to detect likely success or failure after a single dose of MTX with a high degree of certainty. Results showed that if the serum hCG level on day 5 was <200> 60% day 1–5, there seemed to be a high chance of failure. The five cases with a >60% increase in hCG day 1–5 had an unsuccessful outcome with single-dose MTX. Twenty-six per cent of all those with an unsuccessful outcome fulfilled this criterion.

Discussion
TopAbstractIntroductionMethodsResultsDiscussionReferences To the best of our knowledge, this is the first study that has re-evaluated this commonly used protocol for the use of single-dose systemic MTX developed by Stovall and Ling (1993). We believe that the sample size, including women with PULs and those with PTT, reflects the population of women needing MTX likely to be encountered in an Early Pregnancy Unit. In Stovall’s papers, 94.2 and 93.3% of cases of ectopic pregnancy were visualized on ultrasonography (Stovall et al., 1991; Stovall and Ling, 1993). In our study, 68.1% (47/69) of cases were ectopic pregnancies visualized using ultrasonography, but our results show that there was no significant difference in how the protocol worked between those with visualized ectopic masses and persistent PULs.
Many studies have already looked at factors that may help predict success before MTX treatment is initiated. These factors include previous obstetric history, ultrasound findings, folic acid levels, initial serum hCG and progesterone levels and the trend in hCG levels pre-MTX (Saraj et al., 1998; Lipscomb et al., 1999; Tawfiq et al., 2000; Takacs et al., 2004; Gamzu et al., 2002; Dudley et al., 2004; Lipscomb et al., 2004; Takacs et al., 2005). Only a few studies have looked at further possible prognostic variables once MTX has been given. The rates of tubal rupture and trophoblastic persistence have been shown to be higher when there is an immediate increase in hCG post MTX (Dudley et al., 2004; Natale et al., 2004).
This study has shown that Stovall’s original protocol does successfully allow prediction of likely overall success from MTX. Prediction of failure after a single dose of MTX is more difficult to assess, as it is already pre-defined as an hCG decrease <15% href="http://humrep.oxfordjournals.org/cgi/content/full/22/3/858#B2">). We were aware that basing the definition of treatment failure on Stovall’s <15% rule rendered it impossible to comment on the value of the rule for predicting failure. However, we considered the clinical risk of withholding treatment too great to consider it an ethical option. We recognize that others may disagree and argue that only a prospective randomized trial would be able to prove whether a second dose is necessary.
Seven women had a successful outcome without a documented >15% decrease day 4–7. Five of them failed to have hCG levels taken on day 4, but all had a decrease >15% in hCG levels day 5–7. Two women had a suboptimal decrease in hCG day 4–7 (8 and 13%) but refused a second dose of MTX. Both went on to have a successful outcome after just the one dose. It is clear that there can be a successful outcome from a single dose of MTX without a documented >15% decrease in hCG day 4–7, but it would appear that smaller decreases in hCG are associated with an increased risk of failure. It may be possible to redefine the percentage decrease in hCG needed to predict success or failure, but there will be difficulties with variations in hCG assays. Further studies to evaluate the monitoring of success of MTX treatment will need to address this issue.
The second part of this study was to generate new rules, which could predict the likely success of MTX before day 7. However, this study failed to generate a new rule that can supersede current protocols for the medical management of ectopic pregnancy. Rules developed were based on looking for small changes in hCG level between days 4 and 5. Whilst they may allow an earlier prediction of likely success, they are unlikely to work in clinical practice because of the standard error of the serum hCG assay. Logistic regression models have an accuracy for predicting likely success that is closest to that of the day 4–7 rule. However, logistic regression models should be tested on new data before drawing any strong conclusions.
This study has shown that if the serum hCG is <200>60% days 1–5 is associated with a high risk of treatment failure. This could be useful in clinical practice. For those women with an hCG level of <200>60% increase in hCG day 1–5 could be given a second dose of MTX on day 5.
In conclusion, we were unable to develop new rules to predict earlier success of single-dose MTX treatment. We have validated Stovall’s original premise that a 15% decrease in serum hCG between day 4–7 rule is a very good indicator of the likely success of MTX treatment, and consequently, we should continue to use this protocol in the medical management of ectopic pregnancies.

References
TopAbstractIntroductionMethodsResultsDiscussionReferences Condous G, Okaro E, Khalid A, Timmerman D, Lu C, Zhou Y, Van Huffel S, Bourne T. (2004) The use of a new logistic regression model for predicting the outcome of pregnancies of unknown location. Hum Reprod 19:1900–1910.[Abstract/Free Full Text]
Dudley PS, Heard MJ, Sangi-Haghpeykar H, Carson SA, Buster JE. (2004) Characterizing ectopic pregnancies that rupture despite treatment with methotrexate. Fertil Steril 82:1374–1378.[CrossRef][ISI][Medline]
Gamzu R, Almog B, Levin Y, Avni A, Jaffa A, Lessing JB, Baram A. (2002) Efficacy of methotrexate treatment in extrauterine pregnancies defined by stable or increasing human chorionic gonadotrophin concentrations. Fertil Steril 77:761–765.[CrossRef][ISI][Medline]
Hanley JA and McNeil B. (1982) The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 143:29–36.[Abstract/Free Full Text]
Hosmer DW and Lemeshow S. (2000) Applied Logistic Regression(Wiley, New York.).
Lipscomb GH, McCord ML, Stovall TG, Huff G, Portera SG, Ling FW. (1999) Predictors of success of methotrexate treatment in women with tubal ectopic pregnancies. N Engl J Med 341:1974–1978.[Abstract/Free Full Text]
Lipscomb GH, Givens VA, Meyer NL, Bran D. (2004) Previous ectopic pregnancy as a predictor of failure of systemic methotrexate therapy. Fertil Steril 81:1221–1224.[CrossRef][ISI][Medline]
Lipscomb GH, Givens VM, Meyer NL, Bran D. (2005) Comparison of multidose and single-dose methotrexate protocols for the treatment of ectopic pregnancy. Am J Obstet Gynecol 192:1847–1848.[CrossRef]
Natale A, Candiani M, Barbieri M, Calia C, Odorizzi MP, Busacca M. (2004) Pre- and post-treatment patterns of human chorionic gonadotrophin for early detection of persistence after a single dose of methotrexate for ectopic pregnancy. Eur J Obstet Gynecol Reprod Biol 117:87–92.[CrossRef][ISI][Medline]
Saraj AJ, Wilcox JG, Najmabadi S, Stein SM, Johnson MB, Paulson RJ. (1998) Resolution of hormonal markers of ectopic gestation: a randomised trial comparing single-dose intra-muscular methotrexate with salpingostomy. Obstet Gynecol 92:989–994.[Abstract]
Stovall TG and Ling FW. (1993) Single-dose methotrexate: an expanded clinical trial. Am J Obstet Gynecol 168:1759–1765.[ISI][Medline]
Stovall TG, Ling FW, Gray LA. (1991) Single-dose methotrexate for treatment of ectopic pregnancy. Obstet Gynecol 77:754–757.[Abstract/Free Full Text]
Takacs P, Rodriguez L, Laibl V, Pietro P, Kang J. (2004) Relationship of high pre-treatment folic acid level and failure of methotrexate in ectopic pregnancy: pilot study. J Reprod Med 49:713–716.[ISI][Medline]
Takacs P, Chakhtoura N, De Santis T, Verma U. (2005) Evaluation of the relationship between endometrial thickness and failure of single-dose methotrexate in ectopic pregnancy. Arch Gynecol Obstet 6:1–4.[Free Full Text]
Tawfiq A, Agameya AF, Claman P. (2000) Predictors of treatment failure for ectopic pregnancy treated with single-dose methotrexate. Fertil Steril 74:877–880.[CrossRef][ISI][Medline]
Submitted on June 24, 2006; resubmitted on September 10, 2006; accepted on September 14, 2006.
CiteULike Connotea Del.icio.us What's this?

more