An evaluation of language in brain tumour patients using a new cognitive-motivated testing protocol

In patients undergoing tumour resection surgery, assessment of language is vital, given its crucial role in everyday social functioning. However, despite the unique neuropathological mechanisms in tumours, current literature presents variable results regarding language capabilities in this population. In this thesis we have developed a new neuropsychological test battery, the Brief Language Assessment for Surgical Tumours (BLAST), to specifically evaluate language in brain tumour patients. The BLAST adopts a core skills approach, which identifies and examines 11 core cognitive skills that have been derived based on current cognitive and psycholinguistic theories, and are required for everyday language processing. In this study, we administered the BLAST to a cohort of 40 undifferentiated tumour surgery patients, both pre and postoperatively.  Also tested were 60 healthy controls categorised into three age groups (18-29, 30-50 and 51+years). We examined various aspects of overall test performance in order to evaluate: 1) the overall sensitivity of the test battery at detecting abnormalities in this population; 2) selectivity: the relative incidence of impairments across the various subtests; and 3) their sensitivity to change following surgery. We also explored the effects of lesion localisation and other lesion characteristics (malignancy, oedema and volume) on test performance. Following this, we then used participants' test performance to create operationalised measures of our 11 core cognitive skills, and evaluated these measures in a similar way to the basic test scores. Finally, we used Voxel-Based Lesion Symptom Mapping to determine the specific anatomical predictors for each core cognitive skill score. When investigating overall task performance, we found that 94% of preoperative patients and 90% of postoperative patients were impaired in at least one task within the BLAST. Also, 65% and 68% of patients had impaired scores on at least one core skill preoperatively and postoperatively respectively. It was also found that the core skills measures were effective at discriminating amongst different neurological profiles. Specifically, patients with a left posterior tumour had significantly lower scores than other groups on measures of accessing semantic knowledge, lexical selection and phonological encoding, either pre or postoperatively, or both. Conversely, patients with a left frontal tumour had significantly lower scores on measures of articulatory motor planning and verb retrieval. Our Voxel-Lesion-Symptom-Mapping analysis corroborated these findings. Lesions within the left superior temporal lobe significantly predicted lows scores in accessing semantic knowledge, lexical selection and phonological encoding. Conversely, lesions within the left inferior, as well as the superior posterior frontal lobe, significantly predicted low scores on goal-driven response selection, articulatory-motor planning and verb retrieval.  We conclude that a core skills approach may be a more effective means of assessing language in tumour populations than more conventional tools that emphasise overall task performance. Such derived measures are sensitive to impairments in this population, and are less likely to be confounded by nonlinguistic impairments that can impact significantly on overall task scores. They may also be useful in guiding postoperative rehabilitation. Further, the scores derived here are associated with quite specific neural substrates, making them potentially useful in guiding surgery and reducing postoperative linguistic deficits. Finally, we conclude that the investigation of tumour populations can also provide unique theoretical insights into language processing and its neural underpinnings in its own right.