Professional Documents
Culture Documents
org
OBSTETRICS
Ultrasound predictors of placental invasion:
the Placenta Accreta Index
Martha W. F. Rac, MD; Jodi S. Dashe, MD; C. Edward Wells, MD; Elysia Moschos, MD;
Donald D. McIntire, PhD; Diane M. Twickler, MD
OBJECTIVE: We sought to apply a standardized evaluation of ultra- RESULTS: Of 184 gravidas who met inclusion criteria, 54 (29%) had
sound parameters for the prediction of placental invasion in a high-risk invasion confirmed on hysterectomy specimen. All sonographic pa-
population. rameters were associated with placental invasion (P < .001). Con-
structing a receiver operating characteristic curve, the combination of
STUDY DESIGN: This was a retrospective review of gravidas with 1
smallest sagittal myometrial thickness, lacunae, and bridging vessels,
prior cesarean delivery who received an ultrasound diagnosis of
in addition to number of cesarean deliveries and placental location,
placenta previa or low-lying placenta in the third trimester at our
yielded an area under the curve of 0.87 (95% confidence interval,
institution from 1997 through 2011. Sonographic images were
0.80e0.95). Using logistic regression, a predictive equation was
reviewed by an investigator blinded to pregnancy outcome and so-
generated, termed the “Placenta Accreta Index.” Each parameter was
nography reports. Parameters assessed included loss of retroplacental
weighted to create a 9-point scale in which a score of 0-9 provided a
clear zone, irregularity and width of uterine-bladder interface, smallest
probability of invasion that ranged from 2e96%, respectively.
myometrial thickness, presence of lacunar spaces, and bridging
vessels. Diagnosis of placental invasion was based on histologic CONCLUSION: Assignment of the Placenta Accreta Index may be helpful
confirmation. Statistical analyses were performed using linear logistic in predicting individual patient risk for morbidly adherent placenta.
regression and multiparametric analyses to generate a predictive
equation evaluated using a receiver operating characteristic curve. Key words: accreta, invasion, ultrasound
Cite this article as: Rac MWF, Dashe JS, Wells CE, et al. Ultrasound predictors of placental invasion: the Placenta Accreta Index. Am J Obstet Gynecol 2015;212:343.e1-7.
Rac. Placenta Accreta Index. Am J Obstet Gynecol 2015. Rac. Placenta Accreta Index. Am J Obstet Gynecol 2015.
Using variable selection by stepwise desires future fertility. It may also play a
analysis, parameters most highly associ- role in identifying women who may
ated with placental invasion in our benefit from referral to a tertiary center
AUC, area under curve; CI, confidence interval. model are presented in descending order that has sufficient blood bank capacity
Rac. Placenta Accreta Index. Am J Obstet Gynecol 2015. in Table 3. Of the 54 women with his- and multidisciplinary services. Also,
tologic confirmation of placental instead of using each ultrasound
FIGURE 8 TABLE 4
Receiver operator curves for prediction of placental invasion Value of each parameter is
added together to generate
Placenta Accreta Index score
Parametera Value
2 cesarean deliveries 3.0
Lacunae
Grade 3 3.5
Grade 2 1.0
Sagittal smallest myometrial
thicknessb
1 mm 1.0
<1 but 3 mm 0.5
>3 but 5 mm 0.25
Anterior placenta previac 1.0
Bridging vessels 0.5
a
If parameter is not present, then value is 0; b Meas-
ured in sagittal plane; c If any portion of placenta is
anterior.
Rac. Placenta Accreta Index. Am J Obstet Gynecol
2015.
TABLE 5
Sensitivity, specificity, and positive and negative predictive values at each PAI score
Probability of Sensitivity Specificity PPV NPV
PAI n invasion, % (95% CI) (95% CI) (95% CI) (95% CI) (95% CI)
>0 1 5 (1e15) 100 (88e100) 19 (10e31) 38 (27e49) 100 (72e100)
>1 1 10 (4e22) 97 (82e100) 47 (34e61) 47 (34e61) 97 (82e100)
>2 2 19 (10e32) 93 (77e99) 58 (44e70) 52 (38e66) 94 (81e99)
>3 4 33 (22e47) 86 (68e96) 68 (54e79) 57 (41e72) 91 (78e97)
>4 6 51 (36e66) 72 (53e87) 85 (73e93) 70 (51e85) 86 (75e94)
>5 6 69 (50e83) 52 (33e71) 92 (81e97) 75 (51e91) 79 (68e88)
>6 2 83 (63e93) 31 (15e51) 100 (94e100) 100 (66e100) 75 (64e84)
>7 2 91 (73e97) 24 (10e44) 100 (94e100) 100 (59e100) 73 (62e82)
>8 5 96 (81e99) 17 (6e36) 100 (94e100) 100 (48e100) 71 (60e81)
CI, confidence interval; NPV, negative predictive value; PAI, Placenta Accreta Index; PPV, positive predictive value.
Rac. Placenta Accreta Index. Am J Obstet Gynecol 2015.
placental invasion has previously been Hypervascularity of the uterine-bladder prior cesarean delivery (0 points), ante-
described.6,11,12,21,22 Finberg and Wil- interface using both 2- and 3-dimen- rior placenta previa (1 point), grade-1
liams11 found that number and bizarre sional Doppler ultrasound had the best lacunae (0 points), smallest myometrial
appearance of lacunar spaces was diagnostic performance in a recent pro- thickness of 1 mm (1 point), and no
directly correlated with certainty and spective series of women with a prior bridging vessels (0 points) would have a
severity of morbidly adherent placenta. cesarean delivery and placenta previa.9 PAI of only 2, conferring a 10% proba-
In a recent systematic review, overall Similarly, D’Antonio et al17 reported bility of invasion (95% CI, 4e22%).
pooled sensitivity and specificity of that the presence of bridging vessels Depending on availability of local re-
lacunar spaces from 13 studies was 77% identified with color Doppler imaging sources and multidisciplinary care, re-
and 95% with an overall diagnostic ac- was the most predictive ultrasound sults from our index could thus be used
curacy of 0.884.17 A similar relationship parameter of morbidly adherent for counseling and assist with referral
has been shown for number of prior placenta, with an overall diagnostic ac- decisions.
cesarean deliveries.2,18,23 In one of the curacy of 0.95. Limitations of our study include the
largest prospective series to date, Silver Assigning the PAI in clinical practice retrospective design. We were unable to
et al18 found that in the setting of may be helpful in interpreting these control for suspicion of invasion or im-
placenta previa, the risk of accreta was various sonographic variables in light of ages recorded, which leads to certain
11%, 40%, and >60% in women with 1, the patient’s history. As an example, if a ultrasound parameters not being recor-
2, and 3 prior cesarean deliveries, woman who has had 2 prior cesarean ded. As a result, we limited the logistic
respectively. deliveries is found to have an anterior regression to those patients with all pa-
Other characteristics found to be placenta previa with grade-3 lacunae, rameters measured (n ¼ 88) to avoid
positively associated with morbidly smallest myometrial thickness of 1 mm, overfitting our model. Not all women in
adherent placenta in our model, and no bridging vessels, she would our cohort underwent transvaginal im-
although to a lesser degree, include receive 3 points for the prior cesarean aging. Although no studies have been
placental location, grade-2 lacunar deliveries, 1 point for anterior previa, 3.5 performed that directly compare the
spaces, smallest myometrial thickness, points for grade-2 lacunae, and 0.5 diagnostic accuracy of transabdominal
and bridging vessels. Both retrospective points for the smallest myometrial vs transvaginal ultrasound in the setting
and prospective studies,9,11,12,14,16,24 as thickness of 1 mm. This would result in a of suspected placental invasion, trans-
well as metaanalyses and systematic re- PAI score of 8. Her probability of inva- vaginal ultrasound allows for a more
views,17 have shown high sensitivities for sion would be 91% (95% CI 73e97%), complete evaluation of the lower uterine
each of the above parameters. In 2000, with a sensitivity of 24%, specificity of segment and is the current recom-
Twickler et al12 found that a smallest 100%, and PPV and NPV of 100% and mended standard of care.8,25-27 There-
myometrial thickness <1 mm identified 73%, respectively. This is much higher fore, inclusion of transabdominal
in third-trimester pregnancies at risk for than her risk of invasion based on imaging could theoretically affect the
placental invasion was 100% sensitive number of cesarean deliveries and predictive value of our model. Addi-
and 72% specific with a PPV and NPV of placental location alone, which would be tionally, our model was developed only
72% and 100%, respectively.12 only 40%.18 Conversely, a woman with 1 for pregnancies in the third trimester