Interobserver and intraobserver variability in the diagnosis of hydatidiform mole

Surgical pathologists often encounter hydropic villi in products of conception at the first trimester and must determine whether the villi represent complete hydatidiform mole (CM), partial hydatidiform mole (PM), or hydropic abortion (HA). The distinction between these is important for determining the appropriate treatment of patients. This study assessed interobserver and intraobserver variability in the histologic diagnosis of hydatidiform mole among 5 placental pathologists. To evaluate interobserver variability, one representative slide from each of 50 mixed cases of PM, CM, and HA of the first trimester were circulated among 5 placental pathologists. All pathologists used the same histologic criteria by Szulman and Surti. For the second round, the same cases were submitted with DNA ploidy data. For the third round, the slides were recoded and distributed to assess intraobserver agreement. Kappa (kappa) value was calculated for the interobserver agreement in the first and second rounds. There was agreement among 4 or 5 pathologists for only 30 of 50 cases in the first round. There were problems in differentiating between PM and HA in most of the remaining 20 cases. The kappa values varied from poor (kappa = -0.104) to excellent (kappa = 0.761) in the first round. In the second round, there was agreement in 39 of 50 cases and the level of agreement remarkably increased, ranging from fair to good (kappa = 0.552) to excellent (kappa = 0.851). The number of discrepant cases, PM versus HA, was reduced to 4. In 7 cases, there were difficulties in distinguishing CM from HA. The intraobserver agreement ranged from 50% to 90%. Poor interobserver agreement was demonstrated when histology alone was used for diagnosis. Discordance was most frequently seen in PM versus HA and resulted from difficulty in evaluating trophoblastic hyperplasia. Polar trophoblastic growth seen in HA could also be observed in PM. The addition of ploidy data resulted in a significant improvement in concordance. Ploidy study is useful in equivocal cases. Significant interobserver and intraobserver variability was observed even among placental pathologists. New histologic criteria adaptable to differentiation of early lesions are needed.