Presence of a Yolk Sac on Transvaginal Sonography Is the Most Reliable Predictor of Single-Dose Methotrexate Treatment Failure in Ectopic Pregnancy

Transcription

1 Article Presence of a Yolk Sac on Transvaginal Sonography Is the Most Reliable Predictor of Single-Dose Methotrexate Treatment Failure in Ectopic Pregnancy Sarah Bixby, MD, Richard Tello, MD, Ewa Kuligowska, MD Objective. The purpose of this study was to determine which imaging characteristics can be used as prognostic indicators in conjunction with β-human chorionic gonadotropin (β-hcg) levels in the treatment of ectopic pregnancy (EP) with single-dose methotrexate (MTX). Methods. A retrospective study was performed on 62 patients (age range, years; mean, 29 years) treated with MTX for EP from November 2000 to August The transvaginal sonographic findings in each case were analyzed for the presence and size of an extraovarian mass or a pseudogestational sac, amount of free fluid, presence of a yolk sac, and fetal heart motion. Patient age and β-hcg level were also noted. Success of treatment was defined as a single dose of MTX that resulted in appropriate lowering of β-hcg levels. Results. Of 62 patients, 17 (27%) had single-dose MTX treatment failure. A yolk sac was identified in 15 (88%) of the 17 treatment failures and in none of the cases in which treatment was successful (positive predictive value, 100%). The average β-hcg level in the cohort of patients who had single-dose treatment failure was 3282 miu/ml compared with 1544 miu/ml in the treatment success cohort. The presence of fetal heart motion was seen in only 1 patient, and this patient had treatment failure. The age of the patient, size of the extraovarian mass, presence of a pseudogestational sac, and amount of free fluid did not correlate with outcome. Conclusions. The presence of a yolk sac was always associated with treatment failure in single-dose MTX treatment of EP and was the most reliable predictor of failure among all features analyzed. The β-hcg level was a useful adjunct. A prediction rule was created correlating the probability of treatment success with the β-hcg level. Key words: ectopic pregnancy; methotrexate; yolk sac. Abbreviations EP, ectopic pregnancy; hcg, human chorionic gonadotropin; MTX, methotrexate; TVUS, transvaginal ultrasonography Received November 18, 2004, from the Department of Radiology, Boston University Medical Center, Boston, Massachusetts USA. Revision requested November 28, Revised manuscript accepted for publication December 10, We thank Kristen McClure, Waleska Pabon- Ramos, and Linda Mason for help in the preparation of the manuscript. Address correspondence to Sarah Bixby, MD, Department of Radiology, Boston University Medical Center, 1 Boston Medical Center Pl, Boston MA USA. sdbixby@alumni.princeton.edu The diagnosis and management of ectopic pregnancy (EP) is a controversial topic in the fields of radiology, obstetrics and gynecology, and emergency medicine. Ectopic pregnancy represents nearly 2% of all pregnancies and remains the most common cause of maternal death in the first trimester of pregnancy, accounting for nearly 10% of all maternal deaths. 1 The incidence of EP increased considerably from 1970 to 1992 (the last official year for reporting EP), from 4.5 to 19.7 per 1000 pregnancies. 1 In the same interval, the overall fatality rate declined from 35.5 maternal deaths per 10,000 EPs in 1970 to 2.6 deaths per 10,000 EPs in Earlier detection of EP through human chorion by the American Institute of Ultrasound in Medicine J Ultrasound Med 2005; 24: /05/$3.50

2 Methotrexate Treatment Failure in Ectopic Pregnancy ic gonadotropin (hcg) radioimmunoassays and improvement of ultrasound technology with the introduction of transvaginal ultrasonography (TVUS) has contributed to the decrease in maternal mortality associated with the condition. In 1982, methotrexate (MTX) administration for EP was initiated as an alternative to surgical treatment. 2 Methotrexate treatment avoids the morbidity of anesthesia and surgery, allows women to be treated on an outpatient basis, and is more cost-effective. Compared with laparoscopic surgery, MTX therapy saves an average of $3000 per resolved EP 3 and more than $282 million dollars in savings annually. 4 Methotrexate is a folic acid analog and acts as an inhibitor of dihydrofolate reductase, an enzyme integral to DNA synthesis and repair mechanisms. Methotrexate is selective for rapidly dividing cells, such as trophoblastic tissue, and alters the normal development of trophoblastic stem cells, 5 leading to ectopic resorption. The dose is much lower than that used for oncologic purposes; therefore, the side effects are less. Possible side effects of MTX treatment include nausea (20%), oral irritation (8.6%), and diarrhea (2.8%). 6 Hepatotoxicity is a concern with prolonged treatment regimens but is not relevant for the short-term dosing in EP. Success rates for the intramuscular administration of single-dose MTX for treatment of EP range from 64% to 95% as reported in the literature Fertility rates after MTX therapy are 80% compared with 60% after surgical intervention. 14 Currently, MTX is used in common practice for the treatment of EP. The predictors of a favorable outcome, however, are not well defined. In clinical practice, TVUS characteristics are used as exclusionary criteria for MTX therapy. These findings may alert clinicians to the possibility of treatment failure with MTX and may direct patients to a more successful surgical approach. This study evaluated a series of 62 patients over a 3-year period who received MTX for EP and follows outcomes of treatment. The goal of this study was to determine which criteria, either alone or in combination, are the best predictors of treatment outcome with the use of single-dose MTX. Materials and Methods Patients Our Institutional Review Board approved this study. A computer-generated list of pharmacy records of patients treated with MTX between November 2000 and August 2003 was obtained. The computerized medical records were reviewed to identify those patients treated for EP and to ascertain the outcome of treatment. Patients were excluded if TVUS imaging was not performed (or was unavailable for review), if the patients were lost to follow-up, or if suction curettage revealed evidence of intrauterine pregnancy with spontaneous abortion (presence of chorionic villi in the pathologic specimens). Eighty-four patients were treated for EP with MTX. Nine patients were later shown to have undergone spontaneous abortion; 3 patients went to surgery before the MTX was actually administered; 1 patient was transferred to another hospital; 8 patients were lost to follow-up; and 1 patient s ultrasonographic report could not be retrieved for review. A total of 62 patients were eligible for the study (mean age, 29 years; range, years). Transvaginal Ultrasonography Transvaginal ultrasonography was performed on all women seen with a positive β-hcg titer and bleeding, pain, or both. The TVUS was performed by a certified technologist using a 9.5- MHz transvaginal probe (HDI 5000 and HDI 3000; Philips Medical Systems, Bothell, WA). The TVUS images were retrospectively reviewed by 2 radiologists and analyzed according to our protocol. The protocol included multiple parameters: (1) visualization of an extraovarian mass; (2) presence of a yolk sac or fetal pole; (3) presence of fetal heart motion; (4) abnormal fluid collection in the endometrial cavity (pseudogestational sac); and (5) presence and amount of pelvic fluid. Diagnosis of EP Sixty of 62 patients went on to have suction curettage despite the suggested diagnosis of EP on TVUS. Two patients refused suction curettage. A third patient was treated on the basis of results of a saline float test in the Emergency Department, and no specimen was sent for pathologic examination. An exclusionary criterion for suction curettage was visualization of an extrauterine yolk sac or fetal pole on TVUS. In patients with suspected diagnosis of EP by TVUS and β-hcg levels, the diagnosis was confirmed by the absence of chorionic villi in the specimens obtained by suction curettage. 592 J Ultrasound Med 2005; 24:

3 Bixby et al Methotrexate Protocol All 62 patients were treated by the Department of Obstetrics and Gynecology with the single-dose MTX protocol outlined by the Pharmacy and Therapeutics Committee at our institution. This closely followed the original protocol introduced by Stovall and Ling 8 in The indications for receiving MTX include patients who (1) are compliant with follow-up, (2) are hemodynamically stable with no ultrasonographic indication of tubal rupture (ie, no notable free fluid in the pelvis), and (3) have no renal, hematologic, or hepatic disease. A relative contraindication to MTX was a β-hcg level greater than 5000 miu/ml. Absolute contraindications to MTX were the presence of fetal heart motion and an extraovarian mass larger than 3.5 cm. Methotrexate was given intramuscularly at a dose of 50 mg/m 2 of body surface area, which was calculated from a nomogram based on the patient s height and weight (average dose was mg). Additional β-hcg samples were drawn after 4 and 7 days. If the β-hcg level from day 4 to day 7 did not decrease by 15%, the initial dose of MTX was repeated on day 7. If the β-hcg level increased by 50% or more from day 4 to day 7, the patient was admitted for surgical intervention. If the β-hcg level was falling appropriately from day 4 to day 7, additional measurements were acquired weekly until the β-hcg level reached less than 2 miu/ml. Statistical Analysis Continuous variables (age of the patient, β-hcg level, and size of the extraovarian mass) were calculated as a mean value ± SD. Categorical variables (presence of a yolk sac, free fluid, and a pseudogestational sac) were calculated as a percentage. The Student t test was used to measure outcome against continuous variables, and χ 2 analysis was used to measure outcome against categorical variables. All data collected were placed into a spreadsheet (Excel; Microsoft Corporation, Redmond, WA). Univariate analysis was used to determine the significance of each factor in predicting treatment outcome by odds ratio evaluation. The variables significant at the P <.05 cutoff were collected together, and colinearity, confounding, and effect modification were assessed by forward selection, backward selection, and selective elimination. 15 Odds ratios were then calculated for the resulting best fit model with the receiver operating characteristic and Hosmer-Lemelshow goodness-of-fit statistic calculated. 16 With the use of variables significant at the P <.05 level, a prediction rule was constructed on the basis of the logistic regression model technique. 15 Statistical analysis was performed with Stata version 5.0 software (StataCorp, College Station, TX). Results Among the 62 women included in this study, 45 (73%) were successfully treated with single-dose MTX, and the remaining 17 women (27%) had treatment failure. An extraovarian mass was identified at TVUS in all patients; three fourths of the patients had varying amounts of free fluid; one half had a pseudogestational sac; and one fourth had a yolk sac (Table 1). A fetal heart rate was identified in 1 patient. As expected, this patient had single-dose MTX treatment failure because the presence of fetal heart motion is already known to be a contraindication to MTX treatment for EP. According to the protocol outlined above, this patient should not have received MTX. However, she was treated emergently on the basis of an ultrasonographic examination performed several hours earlier at another institution where a fetal pole was documented but no fetal heart motion was shown. Those features that showed a statistically significant correlation with MTX treatment failure were the presence of fetal heart motion (P <.0005), the presence of a yolk sac (P <.0005), and an elevated β-hcg level (P =.004) (Table 2). The size of the Table 1. Characteristics of 62 Women Treated With MTX Patient Characteristics Value Age, y* 29 ± 6 Serum β-hcg, miu/ml* 2037 ± 2070 Size of mass, cm* 2.4 ± 0.8 Pseudogestational sac, n (%) 31 (50) Free fluid, n (%) None 15 (24) Trace/small 29 (47) Moderate 11 (18) Large 7 (11) Yolk sac, n (%) 15 (24) Fetal heart motion, n (%) 1 (1.5) *Values are mean ± SD. Amount of free fluid is defined as follows: trace/small, confined to cul-de-sac; moderate, fluid in cul-de-sac extending into adnexa; and large, fluid in abdomen. J Ultrasound Med 2005; 24:

4 Methotrexate Treatment Failure in Ectopic Pregnancy Table 2. Patient Characteristics Related to Efficacy of Single-Dose MTX Treatment of Ectopic Pregnancy Characteristic Success Failure P Age, y Size of mass, cm Free fluid, n (%) None 8 (18) 7 (41).200 Trace/small 22 (49) 7 (41).200 Moderate 8 (18) 3 (18).200 Large 7 (15) 0 (0) Pseudogestational sac, n (%) 21 (46) 10 (59).361 Serum β-hcg, miu/ml * Yolk sac, n (%) 0 (0) 15 (88) <.005* Fetal heart motion, n (%) 0 (0) 1 (1.5) <.005* Treatment was successful in 45 patients and failed in 17. Free fluid is as defined in Table 1. *Values are statistically significant. mass, presence of pseudogestational sac, and amount of free fluid in the pelvis were not statistically significant factors. Logistic Regression Analysis of univariate significance of the odds ratio showed that significant variables at the P <.05 cutoff were β-hcg level (P =.004), yolk sac (P <.0005), and fetal heart motion (P <.0005). Because fetal heart motion was only seen in 1 patient, this was not included in the focus of our analysis. The adjusted odds ratio for the β-hcg level was /IU, meaning that for every 1-point increase in the β-hcg level, there is a decrease in the odds that the patient will have success with single-dose treatment. Odds ratio analysis for the presence of a yolk sac and fetal heart motion could not be accurately calculated because the number of patients with either of these 2 findings that had treatment success was 0. Age, amount of free fluid, and size of the extraovarian mass were not significant as univariate predictors. The presence of a yolk sac and β-hcg levels were found to be colinear variables, with β-hcg the more accurate predictor. No confounding or effect modification was seen with these factors. Receiver operator curve analysis 16 showed that the use of these factors allows accurate differentiation of a response from no response with an area under the curve of 0.72 (95% confidence interval, ; Figure 1). Model Construction With the use of the variables significant at the P <.05 level a prediction rule was constructed on the basis of the logistic regression model technique to establish the probability of an MTX response given the β-hcg level 15 using the data from this study. Thus, with the β-hcg level, a probability distribution for the probability of a response can be constructed (Figure 2). The probability function is given by the equation exp( β* )/[1 + exp( β* )], where β* stands for the β-hcg level in milli-international units per milliliter. Figure 1. Receiver operating characteristic curve showing performance of a regression model for predicting single-dose MTX treatment failure in patients with EP. The area under the curve is 0.72, suggesting good performance of the model in predicting outcome. The diagonal line represents a test that is no better than chance. Figure 2. Estimated probability of success of single-dose MTX for treatment of EP on the basis of the β-hcg value. A β-hcg value of 1000 miu/ml corresponds to an approximately 80% probability of success. The probability of success falls dramatically as the β-hcg increases: at a β-hcg value of 5000 miu/ml, the probability of success is 50%, and at 10,000 miu/ml, the probability of success falls to less than 20%. 594 J Ultrasound Med 2005; 24:

5 Bixby et al Discussion In this study of 62 patients with EP, the presence of a yolk sac on TVUS was the most reliable predictor of MTX treatment failure (Figures 3 and 4). The β-hcg level was a useful adjunct because higher rates of treatment failure were seen with increasing β-hcg levels. Only 1 patient had fetal heart motion, and this patient had MTX treatment failure. Transvaginal ultrasonographic findings including size of extraovarian mass (Figure 5) and the presence of a pseudogestational sac (Figure 6) did not correlate with treatment outcome. Although a substantial amount of free fluid in the pelvis was considered a contraindication to MTX treatment (because it implies tubal rupture), 7 patients with a large amount of free fluid were treated. Interestingly, all 7 patients had treatment success with singledose MTX. These 7 patients were treated with MTX despite the volume of free fluid because they were deemed stable and had no absolute contraindications to MTX therapy (eg, fetal heart motion or adnexal mass >3.5 cm). The volume of free fluid was assessed subjectively at the time of imaging, and only retrospectively was the volume graded (as outlined in Table 2). Additionally, the amount of free fluid was not always conveyed accurately to the treating physicians, especially when echogenic fluid was present (Figure 7). The fact that all 7 patients had success with treatment is certainly an interesting and unexpected result. However, in our study population, the volume of fluid, as an independent factor, did not show a statistically significant relationship with regard to treatment outcome. It would be worthwhile to determine whether this pattern persists in a larger patient population. On the basis of our results, we suggest that a substantial amount of free fluid in the pelvis in an otherwise stable patient should not be a contraindication to MTX treatment. The complications of failed medical treatment of EP are severe and include tubal rupture and a risk of future infertility. Multiple investigations in the past 2 decades have sought to identity which imaging features, if any, correlate with successful medical treatment and thereby reduce treatment delays and subsequent complications for patients who are not treated successfully. In their study, Lipscomb et al 9,10 found the β-hcg levels the best indicators of treatment success. Fetal heart Figure 3. Yolk sac identified in an adnexal mass in a patient who had single-dose MTX treatment failure. OV indicates ovary; and UT, uterus. motion was also a statistically significant predictor of treatment failure (P =.01). Similarly, Potter et al 11 found that the initial β-hcg value was statistically lower in patients who had successful medical treatment (P <.002). The presence of a yolk sac was identified as a risk factor for failure (P <.02). Potter et al 11 were among the first to identity a yolk sac as an independent predictor of MTX treatment failure in EP. The β-hcg level has been well recognized for its role in identifying patients who are candidates for MTX treatment. Our results substantiated this role by showing that the rate of treatment Figure 4. Extraovarian mass containing a yolk sac in a patient who had single-dose MTX treatment failure. J Ultrasound Med 2005; 24:

6 Methotrexate Treatment Failure in Ectopic Pregnancy Figure 5. Transvaginal sonogram showing an EP appearing as an extraovarian mass. RO indicates right ovary. failure increased with rising β-hcg values according to a prediction rule. Prediction rules are gaining a greater hold in evidence-based medicine and have been used to evaluate problems such as ankle and knee trauma, 17 abdominal pain, 18 and characterization of hepatic lesions. 19,20 We have developed a prediction rule correlating the likelihood of successful single-dose MTX treatment with the β-hcg level, thereby aiding in the clinical management decisions of these patients. Figure 6. Large pseudogestational sac in the endometrial cavity. This patient had successful single-dose treatment. In addition, our data revealed that the presence of a yolk sac on TVUS is specific to those patients likely to have single-dose MTX therapy failure, predicting failure perfectly. The presence of a yolk sac relates colinearly with the β-hcg level, which makes intuitive sense. As the β-hcg level increases, the patient is more likely to have a yolk sac and, therefore, more likely not to respond to single-dose MTX. Lipscomb et al 9,10 postulated that the factors that affect the failure rate of MTX relate to the health of the conceptus, and an EP with a high β-hcg value and fetal heart motion is still developing and growing. We supplement this theory, adding that the presence of a yolk sac is an even earlier indicator of a thriving pregnancy, preceding the presence of fetal heart motion. The presence of a yolk sac heralds that an EP is not likely to respond to single-dose MTX, allowing earlier surgical intervention or anticipation of a multiple-dose regimen. Just as the presence of extrauterine fetal heart motion is commonly considered an exclusion criterion for MTX therapy, the presence of a yolk sac should also be an exclusion criterion for the single-dose protocol. The overall success rate of single-dose MTX therapy at our institution (73%) is lower than that published for other institutions In large part, this is due to the way in which our success rate was calculated. In truth, singledose is oftentimes a misleading term. Patients who are initially enrolled in a single-dose protocol are given more than 1 dose of MTX 14.6% of the time. 21 Women receiving the single-dose protocol can theoretically be given up to 4 doses of MTX and still not be considered part of a multidose regimen, which requires higher doses of MTX and leucovorin rescue. The additional dosing results in increased side effects, longer time to resolution, and potential delay in treatment for those patients who will ultimately require surgery. Our study considers patients successfully treated by the single-dose protocol if they received only 1 dose of MTX for resolution of EP. Our institution is an urban teaching hospital serving an inner-city minority population (44% black, 15% Hispanic, and 28% white). Several patients were lost to follow-up and therefore not included in our study. These patients were likely treated successfully, given that patients in whom treatment failed would 596 J Ultrasound Med 2005; 24:

7 Bixby et al presumably have returned for further care. Our true success rate, therefore, would be slightly higher than our calculated success rate given that we excluded 9 patients lost to follow-up. Suction curettage was performed for definitive diagnosis of EP regardless of the β-hcg levels (which ranged from 69 to 11,132 miu/ml). Because many TVUS examinations were interpreted by the on-call radiology resident overnight in the Emergency Department, suction curettage was encouraged to strengthen diagnostic certainty. In several cases, retrospective review of the studies by an attending radiologist revealed findings that were not noted initially. Accepted practice allows treatment with MTX without suction curettage at all in patients with a β-hcg value greater than a discriminatory level (ie, 2000 miu/ml) when TVUS is suggestive of EP. Women at risk of EP with a β-hcg level greater than 2000 miu/ml and no intrauterine pregnancy identified on TVUS have almost an equal probability of having either an EP or a miscarriage. 22 Treating these women with MTX would falsely increase the success rate. Furthermore, spontaneous resolution of documented EP is common in patients with lower initial β-hcg levels and is reported to occur in 4.9% to 24% of all pregnancies This phenomenon artificially inflates success rates, but, as yet, there is no safe and effective way of identifying these patients who do not require treatment at all. Although our study contains a relatively small patient population, the statistical power is unlikely to be changed by a larger sample size. The results of our probability analysis show how the β-hcg level can be used to identify which patients are appropriate candidates for single-dose MTX treatment. The TVUS imaging features add crucial information regarding the clinical management of these patients because patients with a yolk sac universally had single-dose MTX treatment failure. A larger sample size would certainly strengthen the validity of these observations by showing the persistence of these patterns as well as determining whether and how often a yolk sac is associated with successful singledose treatment. Figure 7. Echogenic free fluid suggesting hemorrhage in the adnexa of a patient who had successful single-dose MTX treatment. References 1. Ectopic Pregnancy United States, MMWR Morb Mortal Wkly Rep 1995; 44: 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: Morlock RJ, Lafata JE, Eisenstein D. Cost-effectiveness of single-dose methotrexate compared with laparoscopic treatment of ectopic pregnancy. Obstet Gynecol 2000; 95: Creinin MD, Washington AE. Cost of ectopic pregnancy management: surgery versus methotrexate. Fertil Steril 1993; 60: DeLoia JA, Stewart-Akers AM, Creinin MD. Effects of methotrexate on trophoblast proliferation and local immune responses. Hum Reprod 1998; 13: Ander DS, Ward KR. Medical management of ectopic pregnancy the role of methotrexate. J Emerg Med 1997; 15: Stika C, Anderson L, Frederiksen M. Single-dose methotrexate for the treatment of ectopic pregnancy: Northwestern Memorial Hospital three-year experience. Am J Obstet Gynecol 1996; 174: J Ultrasound Med 2005; 24:

8 Methotrexate Treatment Failure in Ectopic Pregnancy 8. Stovall TG, Ling FW. Single-dose methotrexate: an expanded clinical trial. Am J Obstet Gynecol 1993; 168: Lipscomb GH, Bran D, McCord ML, Portera JC, Ling FW. Analysis of three hundred fifteen ectopic pregnancies treated with single-dose methotrexate. Am J Obstet Gynecol 1998; 178: Lipscomb GH, McCord ML, Stovall TG, Huff G, Portera SG, Ling FW. Predictors of success of methotrexate treatment in women with tubal ectopic pregnancies. N Engl J Med 1999; 341: Potter MB, Lepine LA, Jamieson DJ. Predictors of success with methotrexate treatment of tubal ectopic pregnancy at Grady Memorial Hospital. Am J Obstet Gynecol 2003; 188: Alshimmiri MM, Al-Saleh EA, Al-Harmi JA, AlSalili MB, Adwani AA, Ibrahim ME. Treatment of ectopic pregnancy with a single intramuscular dose of methotrexate. Arch Gynecol Obstet 2003; 268: Nazac A, Gervaise A, Bouyer J, et al. Predictors of success in methotrexate treatment of women with unruptured tubal pregnancies. Ultrasound Obstet Gynecol 2003; 21: Barnhart KT, Gosman G, Ashby R, Sammel M. The medical management of ectopic pregnancy: a metaanalysis comparing single-dose and multi-dose regimens. Obstet Gynecol 2003; 101: Barnhart KT, Katz I, Hummel A, Gracia CR. Presumed diagnosis of ectopic pregnancy. Obstet Gynecol 2002; 100: Atri M, Bret PB, Tulandi T. Spontaneous resolution of ectopic pregnancy: initial appearance and evolution at transvaginal US. Radiology 1993; 186: Ylostalo P, Cacciatore B, Sjoberg J, Kaariainen M, Tenhunen A, Stenman UH. Expectant management of ectopic pregnancy. Obstet Gynecol 1992; 80: Shalev E, Peleg D, Tsabari A, Romano S, Bustan M. Spontaneous resolution of ectopic tubal pregnancy: natural history. Fertil Steril 1995; 63: Fernandez H, Frydman R. Conservative management of extra-uterine pregnancy. Contracept Fertil Sex 1994; 22: Korhonen J, Stenman UH, Ylostalo P. Serum human chorionic gonadotropin dynamics during spontaneous resolution of ectopic pregnancy. Fertil Steril 1994; 61: Slaughter JL, Grimes DA. Methotrexate therapy: nonsurgical management of ectopic pregnancy. West J Med 1995; 162: Hosmer DW, Lemeshow S. The multiple logistic regression model. In: Applied Logistic Regression. New York, NY: John Wiley & Sons; 1989: Metz CE. ROC methodology in radiologic imaging. Invest Radiol 1986; 21: Tigges S, Pitts S. Introduction to clinical prediction rules for radiologists. AJR Am J Roentgenol 1999; 173: Roth C, Tello R, Sutherland K, Ptak T. Prediction rule for etiology of vague abdominal pain in the emergency room: utility for imaging triage. Invest Radiol 2002; 37: Tello R, Fenlon HM, Gagliano T, decarvalho VL, Yucel EK. Prediction rule for characterization of hepatic lesions revealed on MR imaging. AJR Am J Roentgenol 2001; 176: Carlos RC, Kim HM, Hussain HK, Francis IR, Nghiem HV, Fendrick AM. Developing a prediction rule to assess hepatic malignancy in patients with cirrhosis. AJR Am J Roentgenol 2003; 180: J Ultrasound Med 2005; 24: