SAN ANTONIO – Preoperative ultrasound assessment had an overall sensitivity of only 8% in the detection of positive sentinel lymph nodes among 2,481 patients with node-positive melanoma in the Multicenter Selective Lymphadenectomy Trial II.
"With a sensitivity of only 8% in this large, multicenter clinical trial, ultrasound is clearly not an effective substitute for sentinel node biopsy to detect metastatic disease and provide accurate staging," Dr. John F. Thompson said at a cancer symposium sponsored by the Society of Surgical Oncology.
Dr. John F. Thompson
As a result of the data, preoperative node field ultrasound has now been removed from the Multicenter Selective Lymphadenectomy Trial II (MSLT II) protocol.
The ongoing phase III MSLT II is designed to determine whether the melanoma-specific survival associated with sentinel lymphadenectomy and postoperative monitoring with serial nodal ultrasound is equivalent to that associated with sentinel lymphadenectomy plus complete lymph node dissection in stage IIa to IIIc melanoma patients with sentinel node (SN) metastases detected by histopathologic or molecular techniques.
Ultrasound has proven value in melanoma follow-up, leading some to suggest that preoperative ultrasound examination of regional nodes could provide accurate staging information and render sentinel node biopsy unnecessary.
To date the literature has been confusing. Some studies suggest high (65%) sensitivity (J. Clin. Oncol. 2009;27:4994-5000), while others suggest much lower (24%) sensitivity (J. Clin. Oncol. 2009;27:5614-9), Dr. Thompson said.
He presented data on 2,481 patients enrolled at 29 centers worldwide from December 2004 to March 2010 who underwent preoperative ultrasound assessment of their regional nodes, followed by sentinel node biopsy. Whether ultrasound was performed before or after the preoperative lymphoscintigraphy was at the discretion of the treating clinician. At 19 centers, excised sentinel nodes were examined pathologically with hematoxylin-eosin stain and immunohistochemistry and, if negative, by reverse transcriptase–polymerase chain reaction (RT-PCR) testing.
Sentinel nodes were removed from 2,788 lymph node fields in the 2,481 patients. Histopathology was positive in 578 fields in 554 patients. Among these, ultrasound was true positive in 46 and false negative in 532.
"Hardly an impressive result," said Dr. Thompson, professor of melanoma and surgical oncology at the University of Sydney and executive director of the Melanoma Institute Australia in North Sydney.
Among the 1,927 patients with 2,210 histologically negative lymph node fields, ultrasound was false positive in 52 and true negative in 1,771.
A total of 387 histologically negative, but RT-PCR–positive sentinel nodes were excluded from the ultrasound analysis because they would be expected to be picked up on ultrasound. However, seven of these cases were in fact reported positive on ultrasound, he said.
Of the remaining 2,401 lymph node fields, the overall sensitivity was 8%, specificity 97%, positive predictive value 47%, and negative predictive value 77%. Sensitivity was the highest at 55% in the axilla, and 0% in the popliteal and epitrochlear areas.
There was, of course, a learning curve with the ultrasound technology, Dr. Thompson noted. Sensitivity nearly doubled from 13% for the first 100 cases to 23% for the subsequent 100 cases. Specificity rose modestly from 93% to 95%.
As observed in other studies, the sensitivity of ultrasound increased with Breslow tumor thickness from just 2.6% for tumors 0-1 mm to 12% for those greater than 4 mm. "The positive predictive value at this stage is up around 81%, so you could perhaps make an argument for using ultrasound in thicker tumors," he said.
Finally, tumor burden data were available for 384 histologically positive sentinel nodes. As expected, the median cross-sectional area was larger at 4.80 mm2 for sentinel nodes having ultrasound true-positive results vs. 0.12 mm2 for those having ultrasound false-negative results. Significantly larger tumors were missed by ultrasound in the axilla when compared with the groin (P = .004).
"I think we can say that SN biopsy remains the most accurate method of regional node staging for patients with newly diagnosed melanoma," he concluded. "Sufficient accuracy cannot be achieved by assessing SNs using ultrasound alone."
MSLT II is sponsored by the John Wayne Cancer Institute. Dr. Thompson and his coauthors said they had no relevant financial disclosures.