Intraobserver and interobserver variability in distinguishing between endocervical and endometrial adenocarcinoma on problematic cases of cervical curettings

We conducted a prospective study where 4 pathologists examined patients' problematic cases of cervical curetting for adenocarcinoma to determine whether it is of endocervical or endometrial origin based on 3 parameters: age, morphology, and immunohistochemistry (IHC) panel. Our aims were to evaluate the intraobserver and interobserver variability and to compare the results using those parameters to the final hysterectomy specimens. The value of morphology, morphology+age, and the combined parameters (morphology+age+IHC) in predicting the correct origin of the tumor was evaluated. The intraobserver agreements ranged from fair to almost perfect for each of morphology, morphology+age, and the combined parameters. The interobserver agreements were fair in the first review and ranged from slight to fair in the second review. The agreements between the diagnosis based on morphology, morphology+age, and the combined parameters compared with the final diagnosis on the hysterectomy specimen were slight (kappa=0.137), fair (kappa=0.290), and moderate (kappa=0.497), respectively. We concluded that (i) discriminating between endocervical and endometrial carcinoma is highly subject to intraobserver and interobserver variability. (ii) Surprisingly, this variability is not affected by pathologists' experience. (iii) An IHC panel adds a useful piece of information to predict the tumor origin. Lastly, even though the combination of morphology, age, and IHC is far from perfect in predicting the correct origin of the tumor, it is still the best available method