Sentinel lymph node status in vulval cancer: systematic reviews of test accuracy and decision-analytic model-based economic evaluation

Vulval cancer causes 3-5% of all gynaecological malignancies and requires surgical removal and inguinofemoral lymphadenectomy (IFL). Complications affect > 50% of patients, including groin wound infection, lymphoedema and cellulitis. A sentinel lymph node (SLN) is the first groin node with the highest probability of malignancy. SLN biopsy would be useful if it could accurately identify patients in whom cancer has spread to the groin, without removing all groin nodes. SLNs can be identified by isosulfan blue dye and/or technetium-99 ((99m)Tc) radioactive tracer during lymphoscintigraphy. The blue dye/(99m)Tc procedure only detects SLN, not metastases - this requires histological examination, which can include ultrastaging and staining with conventional haematoxylin and eosin (H&E) or immunohistochemistry. To determine the test accuracy and cost-effectiveness of the SLN biopsy with (99m)Tc and/or blue dye compared with IFL or clinical follow-up for test negatives in vulval cancer, through systematic reviews and economic evaluation. Standard medical databases, including MEDLINE, EMBASE, Science Citation Index and The Cochrane Library, medical search gateways, reference lists of review articles and included studies were searched to January 2011. For accuracy and effectiveness, standard methods were used and reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Searches were to January 2011, with no language restrictions. Meta-analyses were carried out with Meta-Disc version 1.4 (Javier Zamora, Madrid, Spain) for accuracy; none was appropriate for effectiveness. The economic evaluation from a NHS perspective used a decision-tree model in DATA TreeAge Pro Healthcare 2001 (TreeAge Software, Inc., Williamstown, MA, USA). Six options (blue dye with H&E, blue dye with ultrastaging, (99m)Tc with H&E, (99m)Tc with ultrastaging, blue dye/(99m)Tc with H&E, blue dye/(99m)Tc with ultrastaging) were compared with IFL. Deterministic and probabilistic sensitivity analyses were conducted. For accuracy, of the 26 included studies, most evaluated (99m)Tc/blue dye combined. Four studies had clinical follow-up only for test negatives and five had clinical follow-up for all and IFL for test negatives. Numbers with no SLN found were difficult to distinguish from those with negative SLN biopsies. The largest group of 11 studies using (99m)Tc/blue dye, ultrastaging and immunohistochemistry had a pooled sensitivity of 95.6% [95%