The stem asks which CSF leak statement is wrong, so each line must be checked for accuracy. The first claim places the commonest leak at the anterior skull base near the cribriform plate and fovea ethmoidalis. This holds true, since the bone there is paper thin and tightly fused to dura, making it the usual failure point in both spontaneous and post-traumatic leaks.
The second claim relies on beta-2 transferrin. This protein is essentially unique to cerebrospinal fluid, so finding it in a sample of nasal discharge confirms a CSF leak with very high specificity, and the statement is correct. The third claim uses intrathecal fluorescein, injected through a lumbar puncture so that the CSF turns fluorescent and the leak can be pinpointed under the endoscope; this is an accepted localisation technique and is also true.
The remaining option says gadolinium enhanced T1 MRI is the best test to localise the leak. This is the inaccurate statement. Bony defects are demonstrated best by high resolution CT of the skull base, and the fluid pathway by CT cisternography or heavily T2 weighted MR cisternography. Contrast enhanced T1 sequences are not the imaging of choice for tracing a CSF leak. Since the question wants the exception, this is the answer.
\[\boxed{\text{Gadolinium T1 MRI is best for localising the leak (false)}}\]