Accuracy of Ductal Lavage Tested

April 1, 2003

SAN ANTONIO-Ductal lavage was associated with a high false-negative rate in a recent study, according to Seema A. Khan, MD, associate professor of surgical oncology, Lynn Sage Breast Center at Northwestern University. Moreover, about half of the non-fluid-yielding ducts examined in the study contained cancer, she said at the 25th Annual San Antonio Breast Cancer symposium (abstract 25).

SAN ANTONIO—Ductal lavage was associated with a high false-negative rate in a recent study, according to Seema A. Khan, MD, associate professor of surgical oncology, Lynn Sage Breast Center at Northwestern University. Moreover, about half of the non-fluid-yielding ducts examined in the study contained cancer, she said at the 25th Annual San Antonio Breast Cancer symposium (abstract 25).

"Ductal lavage has aroused a lot of enthusiasm as a possible method of breast cancer detection, but before we can provide this as a tool for detecting breast cancer, several questions must be answered," Dr. Khan said. "These include an understanding of the histologic correlates of our cytologic findings from lavage, the sensitivity and specificity of ductal lavage as a detection method, and the frequency of disease in non-fluid-yielding ducts."

The study attempted to answer some of these questions by performing ductal lavage on women scheduled for mastectomy. Patients underwent ductal lavage with saline in the operating room before surgery. If the cytological sample was evaluable for diagnosis, the lavaged duct was injected with a combination of gelatin dye and Omnipaque. If more than one duct was lavaged, different colors of gelatin were used. The breast was then x-rayed to confirm instillation of the dye through the ductal tree. Breast tissue was chilled and sliced at 2- to 3-mm intervals to yield a mean of 155 blocks per specimen, 56 of which contained dye.

The researchers examined 38 breasts, from 35 women. The median age of the women was 49.4 (range, 32 to 87). Thirty-one of the mastectomies were therapeutic—23 invasive cancers and 8 ductal carcinomas in situ (DCIS)—and 7 were prophylactic (one occult invasive cancer with DCIS), Dr. Khan reported. The mean size of the tumors was 3 cm.

Thirty-four of the breasts (89.5%) produced nipple aspirate fluid, and duct cannulation was successful in 32 (84.2%). The mean number of ductal lavages was 1.4 per mastectomy specimen. Twenty-nine breasts (74.4%) were cytologically evaluable, and 24 breasts (61.5%) were histologically evaluable.

Many False Negatives

When the researchers correlated their cytological findings with the histological findings for each breast, they found significant discordance when the cytological diagnosis was "benign or mild atypia," but excellent correlation when the cytological diagnosis was "marked atypia or malignant."

Among 24 designations of benign or mild atypia, findings from ductal lavage produced 4 true negatives and 20 false negatives; 7 of the false negatives involved mild atypia. Among specimens designated as marked atypia or malignant, there were no false positives and five true positives. "If one adds in the non-fluid-yielding ducts and acellular lavaged breasts, the number of true negatives increased to 6, and the number of false negatives increased to 27," Dr. Khan noted.

The sensitivity of the ductal lavage findings was 16% when all 38 breasts were considered, and the specificity was 100%, with a positive predictive value of 100% and a negative predictive value of 18%. When only the 29 cytologically evaluable breasts were considered, the sensitivity of ductal lavage was 19%, and the specificity was 100%, with a positive predictive value of 100%, and a negative predictive value of 18%.

Dye-Containing Ducts Only

The sensitivity of the ductal lavage findings improved when the researchers restricted their analyses to dye-containing ducts only. This analysis yielded eight false negatives (six DCIS and two invasive cancers) and one false positive. The false positive came from a lower inner quadrant lesion, in which a normal duct containing dye and DCIS without dye were present in the same section, Dr. Khan reported.

Analyzed in this manner, the sensitivity increased to 38.5% (with the one false positive) and 43% when the false positive was reclassified. The specificity (with the one false positive) was 93%, with a positive predictive value of 83% and a negative predictive value of 64%. When the false positive was reclassified, the specificity was 100%, with a positive predictive value of 100% and a negative predictive vale of 66%.

Dye was not injected into 10 breasts. The reasons were no nipple aspirate fluid (four breasts), cannulation failure (one breast), acellular lavage sample (two breasts), dye extravasation (two breasts), and miscommunication in the gross room (one breast).

Non-Fluid-Yielding Ducts

The frequency of disease in non-fluid-yielding ducts remains an important issue, Dr. Khan noted. She reported that 12 out of 22 such ducts (55%) did contain cancer.

Based on her group’s findings, Dr. Khan concluded that ductal lavage cytology generally correlates well with the histological findings in the lavaged duct, but the technique’s sensitivity appears poor in the presence of large cancers. In addition, non-fluid-yielding ducts may contain cancer, including DCIS.