Ma. Frantz et Jc. Prolla, CORRELATION OF ENDOSCOPIC CYTOLOGY AND HISTOLOGY IN ESOPHAGEAL CANCER- RESULTS IN PORTO-ALEGRE, RS-BRAZIL, Cytopathology, 7(1), 1996, pp. 38-53
A retrospective study of oesophageal cytopathology at the Hospital de
Clinicas de Porto Alegre (HCPA), RS, Brazil, from 1989 to 1992 was mad
e to assess the sensitivity, specificity, predictive values and accura
cy of endoscopic cytology and biopsy; and study the correlation betwee
n cytopathological and histopathological diagnosis, Specimens from 94
patients were available for review. The final diagnosis was based on s
urgical pathology and follow up. The 81 patients with cancer of the oe
sophagus had the following sex distribution: 64 males and 17 females (
a 3.7-1 ratio). No tumour was found in 13 patients. The following conc
lusions were made: (i) there is excellent correlation between cytology
and histology in oesophageal lesions sampled by endoscopy; (ii) a cor
rect positive cytologic report was obtained in 77 (95%) of the 81 prov
en oesophageal cancers; a false-negative or unsatisfactory result was
given in four patients. A false-positive diagnosis of cancer was not m
ade. There were 13 true-negative reports. These findings result in a s
ensitivity of 95% with 95% confidence intervals (CI) of 90.26-99.74%;
a specificity of 100% (CI of 98.5-100%); a positive predictive value o
f 100% (CI of 99.3-100%); a negative predictive value of 76% (CI of 55
.7-96.3%); (iii) a correct positive histological report was obtained i
n 67 (83%) of the 81 proven oesophageal cancers; a false-negative or u
nsatisfactory result was given in 14 patients. A false-positive diagno
sis of cancer was not made. There were 13 true-negative reports. These
findings result in a sensitivity of 83% with 95% CI of 74.82-91.18%;
a specificity of 100% (CI of 98.5-100%); a positive predictive value o
f 100% (CI of 99.25-100%); a negative predictive value of 48% (CI of 2
9.16-64.84%); (iv) of 81 patients with proven cancer, in 79 (98%) at l
east one of the methods was positive. In only two patients with cancer
were both methods negative. These findings result in a combined sensi
tivity of 98% (CI of 94.92-100%); a specificity of 100% (CI of 98.5-10
0%); a positive predictive value of 100% (CI of 99.31-100%); and a neg
ative predictive value of 87% (CI of 70-100%). Our series confirms the
value of the combined use of cytology and biopsy for the investigatio
n of oesophageal lesions. However, it should be remembered that even w
ith the combined use of cytology and biopsy there are some tumours tha
t will be negative by both procedures: we had only two such cases, con
firming the rarity of such an event.