Safety of methotrexate administration in women with pregnancy of unknown location at high risk of ectopic pregnancy

To evaluate the safety of current guidelines on methotrexate (MTX) administration in women with pregnancy of unknown location (PUL) who are considered to have a high risk of underlying ectopic pregnancy (EP), and to investigate whether implementation of these guidelines would result in inadvertent exposure to MTX of viable intrauterine pregnancies (IUPs).

discussed in the guidelines of the American College of Obstetricians and Gynecologists (ACOG), American Society for Reproductive Medicine (ASRM), Royal College of Obstetricians and Gynaecologists (RCOG) and National Institute for Health and Care Excellence (NICE) were applied to the PUL database.The number of patients eligible to receive MTX and the number with an underlying viable IUP who would be inadvertently prescribed MTX were calculated.
Results A total of 816 women with PUL were reviewed, of whom 724 had complete data and were included in the final analysis.Six patients had persistent PUL and the remaining 718 had a diagnosis of viable IUP, non-viable IUP, EP or failed PUL.According to the ACOG, ASRM, RCOG and NICE guidelines, the rate of MTX administration among patients with PUL would have been 2.76%, 4.56%, 0.41% and 35.36%, respectively.However, no persistent PUL would have received MTX according to the ACOG, ASRM and RCOG protocols (the NICE protocol identified patients with persistent PUL with a sensitivity of 100%), and the majority of MTX treatments were unnecessary because those patients were later classified as having non-viable IUP or failed PUL.Application of ACOG and ASRM guidance could result theoretically in inadvertent MTX administration to women with an underlying viable IUP at a rate of 4.1/1000 (3/724).

Conclusions Current guidelines used to predict high risk of EP in the PUL population lead to inadvertent MTX administration to women with an underlying viable IUP.
These guidelines should be used wisely to ensure that no wanted pregnancy is exposed to MTX.Women with PUL should be monitored carefully, and MTX should be used judiciously when the location of pregnancy is

INTRODUCTION
In early pregnancy, transvaginal ultrasound (TVS) examination is used to assess pregnancy location and viability after a positive result for serum or urinary human chorionic gonadotropin (hCG) 1 .Around 10% of pregnancies cannot be identified as either an intrauterine pregnancy (IUP) or an ectopic pregnancy (EP) on TVS, and these are described as pregnancies of unknown location (PUL) 2,3 .
Having a PUL is a transient state with a large spectrum of outcomes: viable IUP (VIUP), non-viable IUP (NVIUP), EP, failed PUL (FPUL) and persistent PUL (PPUL) 4,5 .Guidelines for PUL management aim to determine the location and viability of the pregnancy, and usually recommend follow-up visits until a definitive diagnosis of viable or non-viable pregnancy is obtained [5][6][7] .EP is associated with a high risk of complications, such as tubal rupture with intra-abdominal hemorrhage 4 , so some physicians may not feel comfortable with expectant management for PUL, fearing a missed or delayed diagnosis of EP.For this reason, various triage protocols have been developed to evaluate the risk of PUL 4,8,9 .
Medical intervention, commonly in the form of methotrexate (MTX), may be considered to treat asymptomatic PUL that is at risk of being an EP [10][11][12][13] .However, prediction of high-risk pregnancy is not error-free.The decision to intervene with MTX may be taken prematurely and could expose a developing IUP to a teratogenic agent, resulting in earlier gestational age at birth, lower birth weight and birth defects 14 .In a survey of Canadian obstetricians and gynecologists, 40% of participants reported cases in which MTX was administered to a suspected EP that was identified subsequently as an IUP 12 .In addition, patients with a low risk of EP could be overtreated and experience unnecessary side effects of MTX, such as headache, dizziness, nausea, vomiting and diarrhea 15 .MTX often results in miscarriage, and it is impossible to determine whether an early pregnancy failure is a destined miscarriage or the result of iatrogenic complications of MTX 16 , so patients could suffer from physical pain and emotional distress.Therefore, all PUL should be managed expectantly, and MTX should not be considered to treat high-risk PUL until an EP is diagnosed on TVS.
This study aimed to evaluate published triage methods for the management of PUL and to determine the safety of MTX administration in PUL considered high-risk by the published criteria.

Data collection
Data were collected from consecutive hemodynamically stable women who attended the early pregnancy unit at Nepean Hospital, Sydney, Australia, between 5 January 2007 and 10 May 2021.All pregnant women presenting at the unit underwent TVS using a Voluson E8 machine (GE Healthcare, Zipf, Austria), equipped with a 5-9-MHz transvaginal probe (RIC5-9-D; GE Medical Systems, Kretztechnik), performed by gynecology fellows who were supervised by the same senior consultant and ultrasound specialist.
If the initial TVS scan demonstrated no signs of an IUP or EP after a positive urinary or serum pregnancy test (serum hCG > 5 IU/L), the woman was classified as having a PUL and was included in the study.Women with any of the following characteristics were excluded from the PUL population and not included in the database: (1)  evidence of an intrauterine sac on the initial TVS scan; (2) EP observed on the initial TVS scan; (3) presence of heterogeneous and irregular tissue in the uterine cavity (indicating retained products of conception, i.e. incomplete miscarriage) on TVS; and (4) a significant hemoperitoneum above the level of the uterine fundus observed on TVS and/or who were hemodynamically unstable and needed surgical management.
Patients who met the inclusion criteria had peripheral blood taken at first presentation to measure the serum hCG level (at 0 h), and this was repeated after 48 h.Women with PUL were followed up with serial serum hCG measurements and TVS until the final clinical outcome was established (VIUP, NVIUP, EP, FPUL or PPUL).

Pregnancy outcome
Intrauterine pregnancy was defined as the presence of a gestational sac with a double decidual sign in the endometrial cavity.Retained products of conception was defined as the presence of heterogeneous tissue in the endometrial cavity with or without color Doppler flow.EP was defined as the presence of a well-defined inhomogeneous mass or a gestational sac with or without a yolk sac anywhere outside the endometrial cavity, most commonly in the adnexa, but also including the ovary, myometrium (interstitium), cervix, Cesarean scar and cornua 13 .PUL that were never identified on TVS, with serum hCG levels falling spontaneously to < 5 IU/L, were classified as FPUL.If serum hCG levels plateaued in the absence of a pregnancy location being determined on repeat ultrasound scans, these women were classified as having a PPUL.Some patients were diagnosed with an EP at a follow-up appointment, and they were managed by surgical intervention, MTX administration or expectant management, according to the Royal College of Obstetricians and Gynaecologists (RCOG) guideline 17 and patient preference.Patient outcomes were extracted from follow-up medical notes, which allowed us to evaluate how EPs were managed.

Prediction of complication risk
Because women with PUL could have an underlying EP, guidelines for management of EP usually also Methotrexate for pregnancy of unknown location discuss the management of PUL, such as those of the National Institute for Health and Care Excellence (NICE) 6 , Society of Obstetricians and Gynaecologists of Canada 5 , Middle-Eastern College of Obstetricians and Gynecologists 7 , American College of Obstetricians and Gynecologists (ACOG) 18 and American Society for Reproductive Medicine (ASRM) 19 .Medical management is considered for PUL because of the risk of EP and associated complications.Therefore, we analyzed the criteria of ACOG, ASRM, NICE and RCOG 6,[17][18][19][20] for triaging PUL according to the risk of EP.They predict the pregnancy outcome based on a single serum hCG measurement and/or the ratio of hCG at 0 h and 48 h (48 h/0 h) in addition to TVS findings.The criteria mentioned in each protocol were listed and applied to the collected database for analysis.
The ACOG guideline discusses the management of EP when a gestational sac is noted ectopically, but also contains a section on management of PUL 18 .It stipulates that, if a pregnancy cannot be identified on TVS, a conservatively high hCG value (e.g. as high as 3500 IU/L) should be applied to evaluate the risk of EP.Therefore, patients with serum hCG > 3500 IU/L and a non-identifiable location of pregnancy on TVS are considered to have a PPUL or high-risk PUL.
The ASRM guideline describes that hCG values may be used to infer the pregnancy outcome when TVS cannot identify definitively an intrauterine or extrauterine gestation 19 , which correlates with the definition of PUL.Therefore, the criteria mentioned in the ASRM guideline were evaluated.It defines the hCG discriminatory zone for a normal IUP as 1500-2500 IU/L, and states that the hCG level in a VIUP is expected to rise by a minimum of 53% every 2 days, based on the 99% CI around the mean value of hCG rise over time 19 .In addition, it states that a decreasing hCG level indicates that the pregnancy is failing.Therefore, when TVS cannot localize the gestational sac inside the uterus, and the hCG level exceeds 2500 IU/L and shows an increasing trend over 48 h but with a ratio of less than 1.53, these women are considered to have a PPUL or high-risk PUL.
The NICE guideline suggests that women should be referred to early pregnancy assessment services for urgent review if the 48 h/0 h hCG ratio is between 0.5 and 1.63 6 , as this group is considered high risk, but no further details are described regarding continuous assessment.
RCOG refers to the NICE guideline for PUL management.Alternatively, RCOG published an evidence-based approach to manage PUL in 2008, in which the pattern of hCG change and its ratio is used to predict pregnancy outcome 20 .According to these criteria, an IUP would cause at least a 35% increase in hCG within 48 h, and a failing PUL would likely result in a drop in hCG by at least 15%.If the change in hCG over 48 h is > 0.85 and < 1.35, patients are deemed to be at risk of EP and need to be rescanned in 1 week.If the hCG level is > 2400 IU/L and the increase in hCG over 48 h remains suboptimal (< 35%) in 1 week's time, these women are considered to have a PPUL or high-risk PUL and medical management should be considered.

Study design
This was a retrospective study.The protocols to predict the risk of EP among PUL patients published by ACOG, ASRM, RCOG and NICE were applied to the collected database to select patients who might be at high risk and for whom MTX treatment could be considered.
Included patients were divided into two groups as per the protocols: those with a low risk of EP who did not need MTX treatment; and those with a high risk of EP for whom MTX treatment could be considered.Patients with a PUL that behaved biochemically like an EP, but for which the location was not confirmed on ultrasound, were eligible for MTX to minimize the risk of subsequent tubal rupture.The frequency of each pregnancy outcome (VIUP, NVIUP, FPUL, EP and PPUL) in these two groups of patients was counted.The following calculations were performed for each guideline to evaluate their performance in predicting the outcome of PUL, especially EP and PPUL, and their safety for MTX administration.
(1) MTX eligibility rate: the percentage of patients with PUL who could have had MTX treatment.(2) Rate of MTX for VIUP: the percentage of PUL patients theoretically eligible for MTX whose pregnancies were eventually located as VIUP.(3) Sensitivity for EP: the percentage of patients with EP who were detected successfully among those with PUL.(4) Sensitivity for PPUL: the percentage of patients with PPUL that were predicted successfully among those with PUL.
These data were entered into a password-protected, de-identified database using Microsoft Excel (2021; Microsoft Corp., Redmond, WA, USA).Ethics approval for this study was obtained from the Human Research Ethics Committee, Nepean Blue Mountains Local Health District, Penrith, Australia (HERC 2020/ETH02180) on 27 October 2020.

Study population
A total of 816 women with PUL were reviewed during the study period.Of those, 60 did not have a hCG value at 48 h and 32 were lost to follow-up.Therefore, 724 women had complete data and were included in the final analysis (Figure 1).
Before the final outcomes had been confirmed, 20 patients were classified as having PPUL at their follow-up Jin et al.
visit.Of those, 13 had successful expectant management, with 12 being diagnosed as FPUL and one as NVIUP, and one case was diagnosed subsequently as EP.These 14 patients were added to the corresponding subgroups of FPUL, NVIUP and EP in our analysis.The other six patients remained as PPUL after multiple visits; they received single MTX and their pregnancies were resolved successfully.Therefore, they were listed as PPUL in Table 1.
Among the outcome subgroups, the patients with VIUP had the highest mean ± SD hCG 48 h/0 h ratio of 2.07 ± 0.92, compared with 1.35 ± 0.91 in patients with NVIUP and 1.11 ± 0.40 in patients with EP (Table 2).The average hCG level in patients with FPUL decreased over 48 h, with a mean ± SD hCG 48 h/0 h ratio of 0.53 ± 0.46.The average hCG level in patients with PPUL increased over 48 h, with a mean ± SD hCG 48 h/0 h ratio of 1.14 ± 0.25.

Evaluation of four protocols
The sensitivity of the four protocols for predicting EP and PPUL, and the rate of MTX use for VIUP, are listed in Tables 1 and 3.
According to the ACOG protocol, 20/724 (2.76%) patients were predicted to be at high risk of EP and MTX treatment could be considered.However, only 1/20 PUL identified by the ACOG protocol was a confirmed EP; therefore, only 1/70 EPs was predicted successfully by the protocol and the sensitivity of predicting EP was as low as 1.43%.Three of the 20 high-risk patients were found to have VIUP as the final clinical outcome, so the rate of inadvertent MTX use for VIUP was 15.00%.In addition, 16/20 patients were later classified as having either FPUL or NVIUP, so no fetal harm would have been caused by MTX administration, but the treatment was unnecessary, so it could have exposed the women to side effects of MTX.No case of PPUL was identified by the ACOG protocol, so its sensitivity in predicting PPUL was 0%.When the ASRM protocol was followed, 33/724 (4.56%) patients were at high risk of EP and MTX treatment could be considered.Only 3/33 had a true EP, so ASRM successfully detected 3/70 EP and the sensitivity of prediction of EP was 4.29%.Three of the 33 high-risk patients were diagnosed with VIUP, so 9.09% would have been exposed accidentally to MTX.The remaining 27 patients were later classified as having either FPUL or NVIUP, so MTX treatment was unnecessary.In addition, no PPUL was identified by the ASRM protocol, giving a sensitivity of 0%.
Based on the NICE guideline, 256/724 (35.36%) patients were deemed high risk for possible EP and needed urgent referral for further assessment.Among these 256 patients, 59 were true cases of EP, so the sensitivity for EP prediction was 84.29%.All six PPUL were identified by the NICE protocol as high risk, so    its sensitivity for PPUL was 100%.Moreover, 174/256 (67.97%) patients who would be defined as high risk ended up with either NVIUP or FPUL.These patients would have received MTX unnecessarily.Seventeen (6.64%) high-risk patients were diagnosed with VIUP and would have received inadvertent MTX.
The RCOG protocol would have resulted in 3/724 (0.41%) patients being treated with MTX.Of those, one was found to have a true EP and the other two failed spontaneously at follow-up visits.Therefore, no VIUP would have been given MTX, but two pregnancies destined to fail would have been overtreated with unnecessary MTX.Furthermore, the protocol was not sensitive in predicting EP or PPUL, with sensitivities as low as 1.43% (1/70) and 0% (0/6), respectively.

Clinical outcome of ectopic pregnancy
The management of 70 patients who were diagnosed with EP is listed in Table 4. Salpingectomy was performed in 21 cases, with a success rate of 100%.The mean ± SD hCG level at first presentation to the clinic in these patients was 1090.8 ± 1309.3 (range, 99-5430) IU/L.Twenty-four patients with EP received MTX treatment.Of those, 23 cases were resolved (hCG range at 0 h, 62-3235 IU/L; hCG 48 h/0 h ratio range, 0.67-4.18),giving a success rate of 95.83%.However, one patient (hCG at 0 h, 590 IU/L; hCG 48 h/0 h ratio, 1.52) experienced abdominal pain following MTX treatment and emergency laparoscopic salpingectomy was performed.The remaining 25 patients with EP were managed expectantly based on shared decision-making.Of those, 21 (84.00%)EPs resolved spontaneously (hCG range at 0 h, 27-1174 IU/L; hCG 48 h/0 h ratio range, 0.33-1.93),but four (16.00%) patients presented with abdominal pain later and underwent laparoscopic salpingectomy (hCG range at 0 h, 440-1867 IU/L; hCG 48 h/0 h ratio range, 1.15-1.56).

DISCUSSION
This is the first study to evaluate established guidelines in the medical management of high-risk PUL.A large Jin et al.
database was analyzed comprising over 10 000 patients seen over 16 years.

Methotrexate treatment in pregnancy of unknown location
In our study, 16.7% of patients with PUL were classified subsequently as VIUP, similar to reported data 5 .Therefore, the risks of overtreating VIUP and causing associated adverse pregnancy outcomes should not be underestimated.
There is no international consensus on the management of PUL, so patients are managed differently in different countries and centers.Serum hCG level can vary enormously between patients due to interpersonal and intrapersonal factors, which was indicated by the large range in hCG level at the patient's first visit and the large SD of the 48 h/0 h hCG ratio (Table 2).The heterogeneity in patient characteristics added further difficulty to predicting pregnancy outcome based on hCG.
In the ACOG guideline, a single hCG measurement is used to predict pregnancy outcome and a high discriminatory level is used (hCG > 3500 IU/L) to minimize the risk of misdiagnosis and overtreatment of VIUP.In contrast, the ASRM and RCOG protocols take both the single hCG measurement and the hCG 48 h/0 h ratio into consideration, albeit with different cut-off values.The relatively high discriminatory level for hCG reduces the risk of misdiagnosing VIUP, but consequently compromises its sensitivity for detecting EP.In this study, the mean ± SD hCG level in patients with EP was 624.5 ± 902.1 IU/L (Table 2), which is much lower than the discriminatory values set by ACOG (3500 IU/L), ASRM (2500 IU/L) and RCOG (2400 IU/L).Consequently, EP with relatively low hCG were excluded by the screening protocols, and the sensitivities of the ACOG, ASRM and RCOG criteria in predicting EP were only 1.43%, 4.29% and 1.43%, respectively (Table 3).The aim of managing PUL with intervention is to minimize the risks associated with a possible underlying EP.Hence, low sensitivity could result in a failure or delay to identify an EP, increasing the risks of mortality and morbidity.Furthermore, high cut-off values for hCG are not a guarantee of accurate prediction of IUP.Other factors should be taken into consideration when applying discriminatory zones, such as technically challenging scans due to distorted anatomy or presence of abnormal anatomy leading to difficulty in detecting a gestational sac.In addition, multiple pregnancy is expected to have a higher hCG level.
Among the guidelines we evaluated, application of the NICE protocol led to the highest percentage of patients falling into the high-risk category for a potential EP.The NICE guideline does not outline any management at this stage and leaves the decision of whether to intervene to the discretion of the clinician.If intervention with MTX is assumed, 6.64% of patients defined as high risk would have incorrectly received MTX for a VIUP.Compared with the other guidelines, the NICE protocol would have caused the highest number of inadvertent MTX exposures to VIUP (17/724).Moreover, 67.97% of patients classified as high risk turned out to be cases of NVIUP or FPUL that were low risk and had the possibility of being overtreated with unnecessary MTX.
If the ACOG and ASRM protocols were applied as definitive selection criteria for intervention, they could potentially result in inadvertent MTX administration to women with VIUP at a rate of 15.00% and 9.09%, respectively (Table 3).The ACOG, ASRM and RCOG protocols resulted in unnecessary treatment of patients with FPUL and NVIUP at a rate of 80.00%, 81.81% and 66.67%, respectively.This concern is discussed in the ASRM guideline because of the potential interruption of a viable pregnancy 19 .
These criteria are not safe enough to use as a standalone triage tool for predicting pregnancy outcome and guiding treatment in PUL, because of the significant risk of mistreatment or overtreatment.Teratogenicity of MTX is not a concern for failing or non-viable pregnancies, but could cause unnecessary side effects, emotional distress to women who experience early pregnancy failure and additional cost to healthcare system.
Therefore, more research is necessary to improve the accuracy of criteria for predicting pregnancy outcome in PUL to achieve higher sensitivities and specificities.Although several multinomial logistic regression models, such as M4 and M6 8,9 , were developed to improve risk prediction in PUL, the misclassification rate could never be zero.To avoid the harm associated with unwanted exposure of pregnant women to MTX, it should not be used in women with PUL whose pregnancy location is unconfirmed.

Transvaginal ultrasound in management of pregnancy of unknown location
Accurate deployment of TVS plays an important role in monitoring and managing PUL.Advanced TVS techniques can reduce the risk of misdiagnosing IUP as PPUL 13 .However, there are several factors that may contribute to an inconclusive TVS scan following a positive hCG test.
First, early dating scans are usually preferred to detect non-viable pregnancy or EP before abnormal clinical presentation 13 , but they are more likely than are scans at later gestational ages to be inconclusive and to classify patients as having a PUL, with the main underlying possibilities including early IUP, occult EP and complete miscarriage 21 .TVS will need to be repeated to monitor the progress of pregnancy alongside serial hCG measurements.
Secondly, the technical skill of the ultrasound operator is associated with the accuracy of the TVS result.Sonographers or sonologists with inadequate scanning or interpretation skills are more likely to fail to locate a pregnancy 22 .There is a learning curve to performing TVS, and the accuracy with which operators detect pregnancy location in the first trimester improves with training 23 .Therefore, ongoing educational programs, quality control monitoring and technique optimization would all improve the competency of sonographers and sonologists for performing TVS and would enable detection of early pregnancy location to ensure the appropriate management.

Conclusions
The current guidelines and protocols to predict the outcome of PUL are not sufficiently conservative to avoid the risk of unwanted MTX exposure.More research is required in this area to improve the accuracy of predicting pregnancy outcome in patients with PUL.In the meantime, monitoring PUL with expert TVS can reduce the risks of misdiagnosis and overtreatment of VIUP.Women with PUL should be monitored continuously and MTX should be used judiciously when the location of pregnancy is not yet confirmed on TVS.

Figure 1
Figure 1Flowchart summarizing inclusion and final pregnancy outcome of patients who were classified with pregnancy of unknown location (PUL) at first clinic visit.*Human chorionic gonadotropin (hCG) values at 0 h and 48 h in addition to transvaginal ultrasound scan.EP, ectopic pregnancy; FPUL, failed pregnancy of unknown location; IUP, intrauterine pregnancy; NVIUP, non-viable intrauterine pregnancy; PPUL, persistent pregnancy of unknown location; VIUP, viable intrauterine pregnancy.

Table 1
Final pregnancy outcome in 724 patients with pregnancy of unknown location, according to risk classification by four protocols 20 and hCG > 2400 IU/L and hCG 48 h/0 h ratio > 0.85 and < 1.35, according to Royal College of Obstetricians and Gynaecologists (RCOG)20. E, ectopic pregnancy; FPUL, failed pregnancy of unknown location; NVIUP, non-viable intrauterine pregnancy; PPUL, persistent pregnancy of unknown location; VIUP, viable intrauterine pregnancy.

Table 2
Human chorionic gonadotropin (hCG) level at presentation and 48 h/0 h ratio in 724 patients with pregnancy of unknown location, according to final pregnancy outcome

Table 3
19aluation of four protocols for predicting risk of ectopic pregnancy (EP) and eligibility for methotrexate (MTX) treatment Data are given as n/N (%).ACOG, American College of Obstetricians and Gynecologists 18 ; ASRM, American Society for Reproductive Medicine19; FPUL, failed pregnancy of unknown location; NICE, National Institute for Health and Care Excellence 6 ; NVIUP, non-viable intrauterine pregnancy; PPUL, persistent pregnancy of unknown location; RCOG, Royal College of Obstetricians and Gynaecologists 20 ; VIUP, viable intrauterine pregnancy.

Table 4
Management and outcome of 70 patients diagnosed with ectopic pregnancy *Data are given as mean ± SD (range), except in case of one datapoint.†Methotrexate.hCG, human chorionic gonadotropin.