CXCL12 predicts relapse in stage I Non-Seminomatous Germ Cell Tumours (NSGCT); results from the MRC TE08 and TE22 clinical trials.


Session type:

Duncan Gilbert1,2, Khin Thway2, Ian Chandler3, Reem Al-Saadi2, Sally Stenning4, Rhian Gabe4, Robert Huddart5,2, Janet Shipley2
1Brighton and Sussex Medical School, Brighton, Sussex, UK, 2The Institute of Cancer Research, Sutton, Surrey, UK, 3Royal Devon and Exeter NHS Foundation Trust, Exeter, Devon, UK, 4Medical Research Council, London, UK, 5Royal Marsden Hospitals NHS Foundation Trust, Sutton, Surrey, UK


Stage I NSGCT harbour occult metastatic disease in up to 50% patients and undergo surveillance with chemotherapy at relapse or adjuvant chemotherapy with excellent survival but concern around treatment toxicities. Better prediction of relapse would minimise exposure to unnecessary treatments. Previous data suggested CXCL12 expression aided prediction beyond current factors (vascular invasion (VI)) in identifying relapse risk[1]. MRC TE08[2] and TE22[3] were clinical trials incorporating surveillance in stage I NSGCT.


Tumour samples from patients enrolled in TE08 and TE22 were obtained with ethical approval and subjected to immunohistochemical staining for CXCL12. These were scored (absent/weak vs moderate/strong) by two independent histopathologists and analysed as a prognostic factor for relapse independent of VI.


Samples were obtained from 190/466 eligible patients, 182 containing tumour sufficient for staining. The 2-year post-orchidectomy relapse-free rate (RFR) was 77.4%; 81.7% (95% CI 74.3, 89.1%) in the 75 with absent/weak CXCL12 intensity and 73.2% (63.3%, 83.1%) in the 107 with moderate/strong. VI was present in ~37% of patients in both groups. CXCL12 was independently prognostic, p=0.034 (logrank test stratified by VI). Using both factors, the 2 year RFR was 94.0% (88.3, 99.7) in the 68 with no VI and moderate/strong intensity, 78.6% (66.8, 90.4) in the 47 with no VI and absent/low intensity, 59.8% (44.1, 75.5) in the 39 with VI and moderate/strong intensity and 51.8% (32.8, 70.8) in the 28 with VI and absent/weak intensity. Absent/weak CXCL12 expression was significantly associated with embryonal (MTU) histology (p=0.001). In multivariate analysis including clinical and pathological features, only CXCl12 (hazard ratio=0.54) and VI (4.36) were predictive of relapse.


CXCL12 intensity is prognostic for relapse independently of VI. It may be used with VI to define groups at low and moderate risk of relapse, but not a group with >50% risk of relapse at 2 years.