The incidence of placenta accreta should increase steadily over the next century as the number of Cesarean sections and maternal age at delivery increase. There is a need for reliable antenatal diagnosis since placenta accreta encountered unexpectedly can lead to catastrophic blood loss, multiple complications such as adult respiratory distress syndrome, Sheehan's syndrome, renal failure, and even death. If these pregnancies can be identified before delivery, the site and time of delivery, as well as the surgical approach, can be planned ahead and blood loss minimized.
This topic is of such importance that all types of imaging have been tried in an attempt to reliably predict placenta accreta with high sensitivity and also a high positive predictive value—underdiagnosing placenta accreta or overdiagnosing it can both create major problems. In the following Review, the term ‘placenta accreta’ includes ‘increta’ and ‘percreta’ since these three entities are usually indistinguishable by imaging.
Risk factors for placenta accreta
In 155 670 deliveries at the University of Southern California, placenta accreta occurred in 9.3% of women with placenta previa and in 0.004% of women without placenta previa. In women with placenta previa, placenta accreta risk varied from 2% in women less than 35 years old with no previous Cesarean section to 39% in women at or over 35 years of age with two or more Cesarean sections. In women with placenta previa, previous Cesarean section and advanced maternal age were independent risk factors. Implantation of the placenta over the scar markedly increased the incidence of placenta accreta1. Note that in the following Review, only papers with cases in which the diagnosis was proven by pathological examination are included. There are many unproven cases in the literature which were identified using ultrasound signs only and no pathological examination was performed.
Placenta accreta (and percreta) does occur in the first trimester. It is usually discovered during dilatation and curettage when massive bleeding occurs due to placental invasion of the myometrium by placenta2–9. This has been, unfortunately, a retrospective diagnosis until recently. Individuals who are at risk for placenta accreta at term are also at risk for placenta accreta in the first trimester, i.e. scarring of the uterus by surgical incision is a risk factor in the first trimester as well as later in pregnancy.
The usual location of a normal early gestation is in the fundus (Figure 1a) or very occasionally in the lower uterine segment (Figure 1b). The sac is surrounded by thick myometrium on all sides. Our group retrospectively reviewed the ultrasound examinations performed up to 10 gestational weeks in women later proven to have placenta accreta on pathological examination10. In the six patients who had scans at 10 weeks or less, all had low-lying gestational sacs, the majority of which were clearly attached to the uterine scar. The myometrium was thin in the area of the scar to which the sac was attached compared to normal in four of the six patients (Figure 2). This low location has also been mentioned in a case report11. There was early fetal demise in two of the six patients, and at the time of dilatation and curettage, bleeding was such that hysterectomy was necessary. In one patient (Figure 2a), there was placenta percreta with downward growth into the cervix. Four pregnancies continued into the second and third trimesters with liveborn infants delivered by Cesarean hysterectomy. Consequently, in these cases, there is hope that the pregnancy will continue but if there is a demise care must be taken to avoid interference with the placenta-uterine interface. This type of pregnancy, in which a sac is abnormally attached in the lower uterus, needs to be differentiated from ‘Cesarean scar pregnancy’ because in the latter, the pregnancy is entirely contained within the myometrial confines of the scar, with no part within the cavity itself.
The low-lying sac attached to the anterior wall of the lower uterine segment needs to be distinguished from a sac that is low lying but surrounded by an equal amount of myometrium, both anteriorly and posteriorly, because sacs may develop into a normal pregnancy (Figure 1b) or they may be in transit to the vagina.
Second and third trimesters
Most patients will first present for ultrasound examination at 18–20 weeks. It is good practice to ask each patient if she has had any uterine surgery, since this increases her risk for placenta accreta. In these cases, the placenta and bladder wall should be carefully examined. Placenta increta with uterine rupture has been reported at this time12–15. We also found ultrasound evidence for placenta accreta at 16–19 weeks' gestation in most patients who were later proven to have it by pathological examination of their uterus.
Although the cause of placental lacunae is unknown, many authors have found them to be predictive of placenta accreta16–19. In 12 of the 14 pathology proven cases we reported who had had a scan at 15–20 weeks, there was at least one finding at the screening scan which suggested placenta accreta20; in 11 of 14 cases this was placental lacunae.
Visualization of lacunae had the highest sensitivity (79%) in the 15–20-week range and a sensitivity of 93% in the 15–40-week gestational age time frame. Three patients who either had no scan during that time or had a scan with no lacunae developed lacunae on subsequent scans. Thus, only one patient with pathologically proven placenta accreta did not have lacunae on ultrasound examination some time during pregnancy. Twickler et al.19 made the additional point that not only did these intraplacental sonolucencies predict all cases of accreta (with three false positives) but also that the lacunae did not need to be near the area of invasion. Finberg and Williams found that the likelihood of placenta accreta increased with the number of lacunae18.
The lacunae give a ‘moth-eaten’ appearance to the placenta (Figure 3) and usually, but not always, have turbulent flow within them, and they appear irregular, often more linear rather than rounded and smooth bordered. They do not have the highly echogenic border that standard venous sinuses have. Tornado-shaped flow of venous, arterial or mixed blood is typical (Figures 4 and 5). These sinuses have been seen as early as 9 weeks' gestation21. Not all large sinuses or vessels are associated with placenta accreta. Large sinuses in patients who did not have placenta accreta can be seen in Figure 6. Note that they are smooth in contour and quite round.
The border between the bladder and myometrium is normally highly echogenic and smooth (Figure 7a). In the case of placenta accreta, interruptions or bulging can occur (Figures 7b and c) and both Finberg and Williams18 and our group found that this is a specific sign, but not a sensitive one, i.e. interruptions or bulging are not present in every patient with placenta accreta. However, this can be a problem in that patients who have had a Cesarean section often develop increased vascularity in the space between the myometrium and the bladder, probably because the bladder flap is retracted before the incision is made into the uterus and because this area is exposed to blood products. Therefore, it would be important to differentiate between bulging due just to enlarged or increased number of vessels and actual growth through the myometrium. Unfortunately, bulging is somewhat non-specific for placenta percreta involving the bladder and does not always predict these cases. In the three cases in our series in which this sign was present, two had a placenta percreta (Figures 7b and c) but one had a simple accreta (Figure 8). Kirkinen et al. noted a bulging of the placenta into the bladder in two of their cases of placenta accreta but there was no growth into the bladder22. Thus, it appears that bulging of the bladder wall may indicate accreta but does not diagnose percreta. Conversely, in the patients we have seen with placenta percreta in which the bladder has been involved, there were abnormalities of the bladder wall on ultrasound inspection. Care must be taken to examine the bladder wall with the ultrasound transducer at 90° so that it is clearly seen (Figure 9a).
Twickler et al. measured the thickness of the lower uterine segment in women who had had a previous Cesarean section and had a low-lying anterior placenta or placenta previa by measuring between the bladder wall and the retroplacental vessels, as seen by color Doppler. All patients later proven to have placenta accreta had a myometrium of less than 1 mm19, which was as predictive of accreta as lacunae. This study has not been repeated to our knowledge.
Loss of the clear space
The usual dark line between the myometrium and the placenta is thought to represent the decidua basalis (Figure 9b). Since the decidua basalis is absent in placenta accreta, it has been suggested that the absence of this line suggests placenta accreta. In fact this line is absent in many normal patients with anterior placentas (Figure 9c)23. We found overall sensitivity of 7% (1/14) for clear space alone at 15–20 weeks and 7% (1/15) from 15–40 weeks. The positive predictive value was 6%.
There are several papers that describe color Doppler in cases of placenta accreta. Unfortunately these papers describe one or two cases and lack any controls. However, Lerner et al. found that turbulent blood flow extending from the placenta into surrounding tissues was very sensitive and correctly identified all patients with accreta. This finding was not present in any of the patients without accreta24. Levine et al., in a blinded study, found that power Doppler did not improve the diagnosis of placenta accreta25. Chou et al. used Doppler in the third trimester and thought that it was reliable. However, on close examination gray-scale images appeared to be as effective and color Doppler could not distinguish accreta from increta26.
Placenta increta is usually treated clinically in the same way as placenta accreta since the surgery and blood loss are the same. However, when the placenta grows through the myometrium (placenta percreta), this growth can be in any direction and can cause significant bleeding and very difficult surgical problems (Figure 10). It is much easier to diagnose placenta accreta than to make the subsequent step of determining if trophoblast has grown through the uterine wall into other structures. Although it would be ideal to identify percreta with certainty, no one has yet been able to do that reliably.
It is important to evaluate more than the bladder wall if we are eventually going to be able to use ultrasound to make this diagnosis since placenta can grow into any surrounding structure in any direction. In one patient seen by us (Figure 10), we had identified probable placenta accreta/increta/percreta. Subsequently, she presented with pain and blood in the abdomen at 26 weeks. Laparotomy showed that the placenta had grown through the uterus into the uterine artery. In another case in our series, the placenta had grown into the broad ligament and vaginal wall. More frequently, the placenta grows through the myometrium to the bladder surface or actually into the bladder wall.
Although adherence of placenta to the bladder wall is not technically invasion of the bladder wall, separation is usually very bloody and can result in a torn bladder wall. This can be very hard to detect since previous surgery in this area makes it quite vascular. Inspection of the bladder wall for integrity is not a foolproof way to determine whether or not the placenta has grown through the wall itself since it can appear interrupted just by vasculature. In a meta-analysis of invasion of the bladder, hematuria was present in only 31% of cases. Cystoscopy was performed in 12 patients and did not make a preoperative diagnosis in any patient. Complications included bladder laceration (20%), urinary fistula (13%), ureteral transection (6%), and a resulting small bladder capacity (4%). Partial cystectomy was performed in 44% of cases. Three maternal deaths occurred in this series27. Even the lack of visualization of the wall can be misleading (Figure 8a). The question clinically is whether or not one needs to know if there is a placenta percreta. The answer is probably ‘yes’ since most obstetricians will be able to handle a Cesarean hysterectomy, but some will want additional help if there is bladder wall invasion.
Placenta accreta without placenta previa or uterine scar
There are patients who have placenta accreta who do not have placenta previa or a history of uterine surgery or other high-risk situations. The placenta may grow through the fundus, for example, in a patient who has never had uterine surgery or instrumentation. These patients may present at birth, but often present earlier with an acute abdomen and copious free blood within it. There are no ultrasound series published as yet that have evaluated the ultrasound appearance of these atypical situations. The present ultrasound literature exclusively addresses the appearance in patients at risk, either with placenta previa or previous uterine surgery or both.
Magnetic resonance imaging (MRI)
Although MRI will probably never be used as a screening tool for placenta accreta, it theoretically should be useful in determining which patients with obvious ultrasound evidence for placenta accreta have placenta percreta, and in confirming placenta accreta in those identified by ultrasound. There have been isolated case reports or small series describing either accreta or percreta, but comparisons with gray-scale ultrasound were not clearly made28–30. In another case, the MRI diagnosis of placenta percreta was not proven adequately by pathology31. In addition, many sequences have imaging times that are long enough for maternal or fetal movement to blur thinned myometrium.
Fortunately, Kim et al. defined the appearance of the placental myometrial interface using the half-Fourier acquisition single-shot turbo spin-echo (HASTE) sequence: it has three layers—inner low signal intensity layer, middle high signal intensity layer and an outer low signal intensity layer. In accreta there was focal non-visualization of the inner layer32. However, this was a retrospective study in which the reviewer knew the diagnosis and all five patients had had a Cesarean section. There are no comparative MRI examinations of the myometrium and placental interfaces and bladder interfaces in patients with placenta previa and previous Cesarean sections and no accreta. When using the criteria of attenuation or non-visualization of the uterine wall in the area of the placenta, interruption of the tissue plane between the myometrium and bladder wall by masses or overt invasion of the myometrium by the placenta, there were five false negatives. Since only pathology-proven cases were considered, there is no estimation of how many false positives there could have been at that institution33. In fact, two of the false negatives had placenta percreta and one placenta increta. Ito et al. found no additional information by MRI in a case of increta, but had used a transurethral ultrasound probe to evaluate the myometrium beforehand, and so had gained as much information as possible from ultrasound34.
Unfortunately there is only one blinded study that addresses the utility of MRI vs ultrasound. In a series of patients with placenta previa and previous surgery, the ultrasound scan, power Doppler and MRI examinations on each patient were reviewed without knowledge of diagnosis or of the results of the other tests. Ultrasound was found to be very accurate and MRI added no additional information in any case except one in which the placenta was posterior (Figure 11)26. MRI will probably be useful in the future when the normal anatomy of this area, as seen on MRI, has been defined and when more data are gathered using fast sequences such as HASTE. Until that time, a high index of suspicion in patients at risk and attention to the cardinal ultrasound signs, particularly irregularly shaped vascular sinuses, will be the standard of care.
There is as yet no completely sensitive and specific test for the diagnosis of placenta accreta. However, when the probability of placenta accreta is raised, a multidisciplinary conference including the patient and family, the anesthesiologist, the imager, and the patient's doctor, will ensure that the entire team understands the issue and the planned clinical approach to the problem. Whether the approach will be a fundal one, with subsequent hysterectomy, or a conservative one in which the placenta is left in place35, blood loss and complications will be reduced if the placenta is not removed from the uterine wall.