Professor, Department of Obstetrics and Gynecology, University of Colorado School of Medicine, Aurora
Dr. Hobbins reports no financial relationships relevant to this field of study.
SYNOPSIS: The authors of this two-center study found that magnetic resonance is not superior to ultrasound in diagnostic accuracy for placenta accreta spectrum disorders, and its usefulness is tempered particularly by a tendency to falsely upgrade the stage of severity.
SOURCE: Einerson BD, Rodriguez CE, Kennedy AM, et al. Magnetic resonance imaging is often misleading when used as an adjunct to ultrasound in the management of placenta accreta spectrum disorders. Am J Obstet Gynecol 2018;218:618.e1-618.e7.
While reading through the January issue of the American Journal of Obstetrics and Gynecology, I came across a letter to the editor1 and response2 pertaining to a paper that appeared last summer in the journal. Although the letters dealt with a mistake in a figure (which had little effect on the overall message of the paper), it re-tweaked my interest in the original paper, which focused on the growing problem of placenta accreta and the efficacy of ultrasound vs. magnetic resonance imaging (MRI) in diagnosing it.
Einerson et al analyzed data generated from two institutions (University of Utah and University of Colorado) over 20 years. Seventy-eight patients suspected of having a possible placenta accreta spectrum (PAS) disorder, often now labeled as morbidly adherent placenta (MAP), were referred for evaluation by ultrasound and MRI. The primary outcome was how often a change in diagnosis could affect clinical management and how often the diagnosis was confirmed clinically at delivery and/or by pathological examination of the hysterectomy specimen.
The diagnoses by both ultrasound and MRI were divided into three categories: 1) no PAS; 2) suspected “mild” PAS, which included accreta or increta; and 3) severe PAS, for which the trophoblastic invasion involved the serosa or tissue outside the uterus (percreta). The management plans were consistent throughout the time span of the study. If PAS was not suspected on MRI, a placental delivery was attempted after vaginal birth (if there was no evidence of placenta previa) or at the time of cesarean delivery (if there was a previa). If there was no suspected PAS, delivery would be accomplished at 37 weeks. If mild PAS was suspected, a cesarean hysterectomy was scheduled for 36 weeks; if severe PAS was suspected, the operation was booked for 34 weeks with specialized personnel in attendance.
The results were as follows:
- MRI changed the diagnosis and management in 28 cases (36%) and confirmed the diagnosis in 64%.
- In the 28 cases of changed diagnosis, MRI was correct in 15 and incorrect in 13.
- In the 50 cases in which MRI agreed with ultrasound, the diagnosis was incorrect in 16 cases (32%).
- MRI overdiagnosed the stage of severity in 23% of cases and underdiagnosed it in 14%.
- The positive predictive value (PPV) of ultrasound in the original paper was 73% vs. 64% for MRI. However, in their reply letter,2 the authors found coding errors in two patients and corrected the PPV to 79% for ultrasound and 61% for MRI, neither of which is ideal.
- There was little difference in detection rates of either modality over the time span of the study, and no differences in accuracies were noted in the few cases of lateral or posterior placentas.
The authors concluded that MRI changed the diagnosis enough to alter clinical management in one-third of cases, “but when changed, it was often incorrect.” They noted that MRI is best used as an adjunct in cases in which ultrasound results are equivocal.
This is certainly not the first paper to assess the value of MRI in patients at risk for PAS. In some papers, MRI comes out with high marks for accuracy of detection. For example, results of two meta-analyses with large numbers of patients found sensitivities for MRI in detecting PAS of 94% and 82%, with specificities of 84% and 88%.3,4 In both studies, MRI was touted as being “highly accurate” in the diagnosis of this condition. However, the authors of the Utah/Colorado study pointed out that the MRIs in these studies were applied to patients already at very high risk for an abnormality because an ultrasound exam had suggested a problem. Since ultrasound most commonly is the first-line test, it could have a pretest probability risk as low as 0.5 and the subsequent MRI would start with the ultrasound post-test probability of at least 0.80 — an unfair advantage in any type of comparison study. Yet in the featured study, MRI was still wrong in 37% of cases.
PAS (MAP) is a strong contributor to maternal mortality and morbidity. The incidence has risen over the last 20 years concomitantly with an increase in the cesarean delivery rate. Although cesarean deliveries have tapered off slightly over the last two years to about 30%, the rate of PAS will not follow this trend because so many patients with two or more previous cesarean deliveries, having a much higher probability of PAS, will enter the at-risk pool.
The key to avoiding a disaster is to suspect it. The bulk of PAS patients are those with previous cesarean deliveries who have a placenta previa, so these patients need ultrasound scans that are highly focused on the following stigmata of PAS:
- Lacunae, especially near the previous uterine incision (these lakes should have documented Doppler flow within them; the risk of PAS increases with the number of the lakes);
- Absence or major thinning of the myometrial clear space under the affected placental interface;
- Hypervascularity in the basal plate of the placenta;
- Irregularity and/or discernible vascularity in the bladder wall;
- I favor the use of three-dimensional directional color Doppler to approach the placental bed en face. With PAS, a chaotic, turbulent blending of vessels is encountered (with typical aliasing) instead of an orderly pattern of discrete maternal and fetal vessels.
Patients at greatest risk for PAS will benefit from an ultrasound exam performed by someone with experience. The quality of ultrasound diagnosis depends on two human steps (acquisition and interpretation). In MRI, the acquisition is handed off to a very complicated and expensive machine with only one human element remaining — interpretation (mostly happening long after the patient is gone). The clues generally are there by the 20th week of gestation, but a second ultrasound evaluation after 30 weeks often can be of further benefit.
MRI is very costly and can be poorly tolerated, especially in claustrophobic patients. The American College of Obstetrics and Gynecology guidelines (2012 and reaffirmed in 2017)5 state that MRI is “an adjunctive modality that adds little to the diagnostic accuracy of ultrasound,” except for possible posterior accreta, for which it might be helpful. My opinion: I think MRI has a diagnostic place in PAS, not as a first-line “go-to” modality, but as an arbiter in some cases in which the ultrasound diagnosis is equivocal. Under these circumstances, an “over call” generally has less potential for disaster than an “under call.”
- Rottenstreich A, Levin G, Zigron R. Magnetic resonance imaging management of placenta accreta spectrum. Am J Obstet Gynecol 2019;220:127.
- Einerson BD, Silver RM, Rodriguez CE. Reply. Am J Obstet Gynecol 2019;220:127.
- D’Antonio F, Iacovella C, Palacios-Jaraquemada J, et al. Prenatal identification of invasive placentation using magnetic resonance imaging: Systematic review and meta-analysis. Ultrasound Obstet Gynecol 2014;44:8-16.
- Meng X, Xie L, Song W, et al. Comparing the diagnostic value of ultrasound and magnetic resonance imaging for placenta accreta: A systematic review and meta-analysis. Ultrasound Med Biol 2013;39:1958-1965.
- Committee on Obstetric Practice. Committee opinion no. 529: Placenta accreta. Obstet Gynecol 2012;120:207-211.