Clinical Endocrinology (1998) 49, 391–396

391q 1998 Blackwell Science Ltd

Cognitive dysfunction in patients with pituitary tumourwho have been treated with transfrontal ortranssphenoidal surgery or medication

K. A. Peace*, S. M. Orme, S. J. Padayatty,H. P. D. Godfrey† and P. E. BelchetzDepartment of Endocrinology, The General Infirmary,Leeds, UK, *Department of Psychological Medicine,Otago University, Dunedin, and †Department ofPsychology, Otago University, Dunedin, New Zealand

(Received 8 July 1997; returned for revision 11 November 1997;finally revised 6 March 1998; accepted 9 April 1998)

Summary

OBJECTIVE This study was carried out to examinethe neuropsychological status of patients treatedfor pituitary tumour by transfrontal surgery, trans-sphenoidal surgery or medical treatment only, withor without radiotherapy.DESIGN AND MEASUREMENTS Three groups of 23patients who had been treated for pituitary tumourwere compared with 23 healthy controls on a range ofneuropsychological measures. The surgical patientswere also subdivided into two groups and compared.The neuropsychological measures were standardizedpsychological tests designed to assess aspects ofattention, memory and executive function.PATIENTS The patients were those who had beentreated with transfrontal surgery ( n ¼ 23), transsphe-noidal surgery ( n ¼ 23) and medication only ( n ¼ 23).The groups did not differ with respect to age,education or premorbid ability level as assessed bythe National Adult Reading Test. All participants werefree of known sources of cognitive impairment otherthan pituitary tumour.RESULTS Comparison of the four groups revealedthat nearly half of the transfrontal, one-third of thetranssphenoidal and one-quarter of the non-surgicalgroup had three or more neuropsychological testsscores below the 10th percentile compared to lessthan 5% of the controls. Impairments in memory andexecutive function were found in both surgicalgroups. The non-surgical patients appeared to have

problems only on tasks requiring high levels of cog-nitive processing. Differences were found betweenthe two surgical groups with respect to the severityof the cognitive impairment, the transfrontal patientshaving more severe impairment than the transsphe-noidal. No significant negative effects on cognitivefunctioning were associated with radiotherapy; how-ever, transfrontal surgery patients who had not beentreated with radiotherapy were found to be moreimpaired than other patients. This was thought to berelated to radical surgery.CONCLUSIONS Many patients with treated pituitarytumour suffer significant cognitive impairment. Theseverity and nature of impairment differs betweentreatment groups, although the cause of this couldnot be addressed by this study. Recommendationsare made for future research and clinical practice.

Although it was widely recognized that tumours of the pituitarygland cause cognitive problems if left untreated, it was not untilvery recently that systematic investigation demonstrated thatpatients continue to suffer impairment in cognitive functionfollowing treatment of their tumour (Grattan-Smithet al., 1992;Peaceet al., 1997). Both the aforementioned studies reportedimpairments in memory and executive function.

Peaceet al. (1997) excluded all patients from their study whohad any known source of cognitive impairment, other thanpituitary tumour, and also demonstrated that the cognitiveimpairment was not secondary to a mood disorder. It wastherefore concluded that the source of cognitive impairmentwas likely to be multifactorial, but may be associated withtreatment variables such as surgery or radiotherapy or due tohormone imbalance resulting from the primary disease orpituitary surgery.

The type of surgery used to remove pituitary tumours haschanged several times in the past. However, the two mostwidely used techniques have been the transfrontal route and thetranssphenoidal routes. The transfrontal route was the mostextensively used until the introduction of intraoperativemicroscopes and better treatment of post-operative infectionmade the transsphenoidal route much safer. Transsphenoidalsurgery offers faster and easier access to the pituitary gland,allows for better differentiation of the tumour from the gland,

Correspondence: Dr K. A. Peace, Department of PsychologicalMedicine, Otago University, Dunedin, New Zealand.Fax:þ64 3474 7934; e-mail: kathryn.peace@stonebow.otago.ac.nz

presents less risk to the optic chiasma and is considered lesstraumatic to the patient. Currently, transfrontal surgery isgenerally used only for large tumours, those with parasellarextensions, or for patients with infections of the sphenoidsinuses or abnormalities of the sella turcica. The effects of thesealternative routes on cognitive functioning are not clear as nodescriptive studies have been carried out which assess relevantvariables. Clinical opinion, however, seems to favour trans-sphenoidal surgery as being less likely to produce damage tothe brain because it does not involve entry to the brain per se, orretraction of parts of the brain. Transfrontal surgery, on theother hand, usually involves retraction of the right frontal lobewhich has been associated with damage to the small perforatingarteries of the internal carotid artery (Horwitz & Rizzoli, 1982)and vasospasm (Mawket al., 1979), both of which can produceareas of infarction.

Radiotherapy has been the subject of study in many groups ofpatients, although not in patients with pituitary tumour wherepsychological variables have been the dependent measures.Radiotherapy is given primarily to inhibit regrowth of tumoursover a long period of time as it is known to reach maximaleffectiveness many years after administration (Sheline, 1974;Flickinger et al., 1989). The appearance of adverse conse-quences affecting the hypothalamus and optic chiasma(Atkinson et al., 1979) and the frontal and temporal lobes(Shelineet al., 1980) approximately 20 years ago led to thereduction in dosage used for pituitary irradiation (Aristizabelet al., 1977). However, parts of the brain still lie within the 60–70% isodense curves and are therefore exposed to considerableirradiation (Bleehan & Glastein, 1983). It remains a possibilitythat radiotherapy for pituitary tumour may lead to adverseneuropsychological consequences.

Radiotherapy, therefore presents a risk factor for many yearsafter administration, especially in patients with pituitary tumouras they have near normal survival rates (Flickingeret al., 1989).Radiotherapy may or may not be carried out following surgerydepending on the confidence of the surgeon that all of thetumour has been removed.

The purpose of the present study was to examine theneuropsychological status of patients treated for pituitarytumour by transfrontal or transsphenoidal surgery, with orwithout radiotherapy. This involved retrospective comparisonof patients who had the two types of surgery with those whoreceived medication only and a control group of healthy adults.The hypotheses for the present study were therefore:

X That patients treated for pituitary disease by surgery will bemore impaired on neuropsychological measures than patientstreated nonsurgically and healthy controls.

X That patients treated with transfrontal surgery will be moreimpaired than those who had transsphenoidal surgery.

X That patients treated with radiotherapy will have greatercognitive impairment than patients who did not haveradiotherapy.

Subjects and methods

Study design

Four groups of patients were evaluated on a range ofneuropsychological measures. The four groups were patientswho had been treated with transfrontal surgery, transsphenoidalsurgery, nonsurgical treatment and healthy adults. The surgicalpatients were then subdivided into those receiving radiotherapyor not and compared.

It is acknowledged at the outset of the study that this was aretrospective study and therefore patients within each treatmentgroup could not be allocated randomly to treatment, as wouldbe preferable from a methodological point of view. Hence con-founding variables such as type of tumour are acknowledgedand the study must be regarded primarily as a description of theneuropsychological characteristics of the various subgroups ofpatients, rather than a comparative study of treatment effectswhere causality of impairment can be clearly attributed.

Ethical committee approval had been given for the study.

Subjects

Twenty-three patients were recruited to each treatment groupfrom Leeds General Infirmary Endocrinology Department. Inaddition, 23 healthy controls were recruited.

All patients had been on optimal and stable hormonereplacement therapy for at least a year. All were growthhormone (GH) deficient and none received GH during thestudy, as is current standard practice. At least 2 years hadelapsed since surgery or radiotherapy in order to be sure thatany acute post-treatment effects had resolved. Patients withsignificant suprasellar extension of the pituitary tumour,hydrocephalus or intraventricular shunts, cerebrovasculardisease and those on antiepileptic medication were excludedfrom the study in order to rule out other known sources ofcognitive impairment. None of the patients were known to haveimpaired vision due to chiasmal damage secondary to radio-therapy, surgery or tumour characteristics.

Of the 46 patients who had been treated surgically, 25 hadalso received radiotherapy treatment. As can be seen fromTable 1, the four groups did not differ with respect to age,gender, years of education or estimated premorbid IQ based onthe NART. There was a difference between the groups withrespect to duration of illness, with the transfrontal patientshaving a considerably longer mean duration of illness(14·9 years, standard deviation (SD)¼ 6·2) than the other

392 K. A. Peace et al.

q 1998 Blackwell Science Ltd,Clinical Endocrinology, 49, 391–396

two groups (transsphenoidal¼ 8·6, SD¼ 6·4 and medical¼ 8·5 years, SD¼ 6·2).

Distribution of types of tumour was also different betweenthe three treatment groups (see Table 2).

Measures

All patients and controls underwent the same battery of testsreported by Peaceet al. (1997) to assess premorbid ability,attention, memory and executive function. The assessment wascarried out in hospital by a psychologist. Rationale for themeasures used is described in the previous publication. Thetests administered were:

Premorbid abilityNational Adult Reading Test(NART) (Nelson, 1982)AttentionDigit span subtest of the Wechsler Adult Intelligence Scale-Revised (Wechsler, 1981)MemoryAuditory–Verbal Learning Test(Rey, 1964)Story recall subtest from the Wechsler Memory Scale(Wechsler, 1945)Recognition Memory Test for Faces(Warrington, 1984)Executive functionsControlled Oral Word Association Test(COWAT) (Benton &Hamsher, 1976)Block Designsubtest of the WAIS-R (Wechsler, 1981)Trail Making Test(Army Individual Test Battery, 1944)

Statistical analysis

All analyses were carried out using an SPSS package forWindows version 6.1. One-way ANOVA was used to com-pare the patients who had received transfrontal or trans-sphenoidal surgery, nonsurgical treatment and controlparticipants. A 2-way ANOVA was used to compare thesurgical patients who had radiotherapy and those who did not.

As the study was primarily exploratory in nature generalrecommendations made by Saville (1990) were followed inorder to minimize Type II errors, namely to use apost-hoctestwhich was not too conservative. Therefore, where a significantdifference was found between the groups, Tukey’s HonestlySignificant Differences was used to explore between groupdifferences.

In order to explore the extent of impairment in the fourtreatment groups the number of subjects in each group whohad three or more test results below the 10th percentileusing standardized normative data was also compared using thex2-square test.

Results

Extent of cognitive impairment

Only 4·3% of the control group had three or more test scoresbelow the 10th percentile, which is consistent with a samplehaving a mean predicted IQ in the above-average range.However, greater numbers of patients in the three treatmentgroups had scores below the 10th percentile. The differences

Cognitive dysfunction in patients with pituitary tumour 393

q 1998 Blackwell Science Ltd,Clinical Endocrinology, 49, 391–396

Table 1 Mean (SD) demographic and premorbid ability data for patients and controls

Transfrontal Transsphenoidal Non-surgical Control

Mean SD Mean SD Mean SD Mean SD P

Age 41·6 13·8 42·7 10·5 38·7 10·6 38·9 13·1 0·6Years of education 11·3 1·6 10·9 0·8 11·5 1·7 11·6 1·8 0·4Duration of illness 14·9 6·2 8·6 6·4 8·5 6·2 — 0·00Estimated premorbid IQ 109·8 10·3 107·9 9·3 111·0 7·1 112·1 7·3 0·4

Females (%) 52 69 73 52 0·2

Table 2 Numbers of patients having differenttypes of tumour in the three patient groups Tumour type Transfrontal Transsphenoidal Non-surgical

Non-functioning adenoma 8 8 6Cushing’s disease 0 9 1Prolactinoma 6 1 14Acromegaly 3 5 1Craniopharyngioma 6 0 1

between the groups were statistically significant (x2 ¼ 8·2;P<0·05), with the transfrontal patients demonstrating thegreatest level of cognitive impairment (43·5%), followed by thetranssphenoidal (30·4%) and the nonsurgical patients havingthe least (21·7%). However, many more of the nonsurgicalpatients had three or more tests below the 10th percentile thanoccurred in the control group, suggesting that many of thesepatients also had cognitive deficits.

Nature of the cognitive impairments

No differences were found between the groups on tests ofattention. However, significant differences were found betweensome of the groups on tests of memory and executive function(see Table 3).

On the memory tests the transfrontal and transsphenoidalpatients had significantly lower scores on all three memorytests, suggesting impairments in both verbal and non-verbalmemory. The nonsurgical group had lower scores only on theimmediate and delayed recall of the story.

On the executive function tests the transsphenoidal patientshad significantly lower scores on the block design test, and thetransfrontal and nonsurgical patients had scores that wereintermediate, but not significantly different to either transsphe-noidal patients or controls. On the other tests of executivefunction there was a trend for the transfrontal patients to havethe lowest scores followed by transsphenoidal patients and thenthe nonsurgical group. However, the results did not reachstatistically significant levels, despite apparently quite largemean differences in test scores. One reason for this, particularlyon the trail making test, may be the considerable variability intest results within some of the groups, particularly thetransfrontal group, suggesting that some patients may beseverely impaired while others are mildly or unimpaired.

Distribution of impaired neuropsychological testperformance

In order to investigate how many patients in each group wereimpaired, and to what extent they were impaired, allneuropsychological test results were assigned a score basedon age-adjusted normative data. Scores were assigned asindicated in an adapted version of Wechsler’s (1981) classifi-cation system as shown in Table 4. For each patient individual testscores were averaged. The mean test score was then compared tothe score for predicted premorbid ability level (i.e. the NARTscore). Level of impairment was defined as follows:

No impairment¼ difference between NART score and mean ofother scores< 1.

Mild impairment¼ difference>1< 1·5.Moderate impairment¼ difference>1·5<2.Severe impairment¼ difference>2.

Table 5 shows the percentage of people in each group withineach of the categories of impairment. These results showed thatin all three of the patient groups there were some patients whowere unimpaired. Differences emerged between the two

394 K. A. Peace et al.

q 1998 Blackwell Science Ltd,Clinical Endocrinology, 49, 391–396

Table 3 Memory and executive function test results

Transfrontal Transsphenoidal Non-surgical ControlTest mean (SD) mean (SD) mean (SD) mean (SD) P Tukey

Story recallImmediate 21·9 (7·3) 18·8 (7·8) 20·3 (8·5) 26·8 (7·6) 0·01 F,TS,NS<CDelayed 17·8 (8·4) 15·4 (8·1) 17·8 (7·7) 23·4 (8·2) 0·01 F,TS,NS<C

ListsTrials 1–5 46·9 (10·2) 50·4 (10·2) 55·7 (7·8) 55·0 (8·6) 0·01 F,TS<NS,CImmediate 9·0 (3·7) 9·9 (3·4) 11·8 (2·4) 11·3 (2·3) 0·01 F,TS<NS,CDelayed 8·0 (3·7) 9·8 (3·4) 11·5 (2·4) 11·2 (2·6) 0·01 F,TS<NS,C

RMT faces 38·4 (4·7) 40·5 (5·4) 42·0 (3·9) 43·6 (4·1) 0·01 F,TS<CBlock design 32·0 (13·1) 28·6 (11·5) 33·7 (9·6) 38·2 (8·4) 0·02 TS<CCOWAT 32·8 (10·4) 34·6 (11·6) 39·4 (11·7) 42·0 (13·1) 0·07Trails B 96·9 (75·0) 87·9 (42·8) 77·4 (50·5) 59·5 (21·8) 0·08

Table 4 Classification for determining mean neuropsychological score

PercentageLabel included Score

Abnormal/borderline 10 1Below average 15 2Low average 25 3High average 25 4Above average 15 5Superior 10 6

surgical groups with respect to level of impairment. In thetransfrontal group patients tended to be either unimpaired orseverely impaired, while in the transsphenoidal group patientswere either unimpaired or mildly impaired, with very fewhaving moderate or severe impairment. In the nonsurgicalgroup the majority were unimpaired, the remainder beingdistributed in all three categories of impairment.

Effects of radiotherapy on neuropsychological testscores

No significant differences were found between the surgicalpatients who had radiotherapy and those who did not.However, there was a significant interaction effect with typeof surgery on the immediate and delayed recall of the listlearning tasks and on part A of the trail making test, with thepatients who had transfrontal surgery but no radiotherapyperforming at a lower level (Table 6). A similar pattern ofscores emerged on many of the other tests but results did notreach significance.

Discussion

The present study demonstrated that many patients treated forpituitary tumour suffer cognitive impairment when compared tohealthy controls. Patients treated surgically are more likely tohave cognitive impairment than those treated nonsurgically.The degree of cognitive impairment which many of the patientsare suffering is of considerable magnitude, such that it islikely to have a significant impact on their capacity tofunction effectively in the workplace, home environment and

interpersonally. This is particularly so for a larger percentage ofthe patients treated with transfrontal surgery.

Comparison of the two surgical groups revealed little in theway of difference on mean test scores, suggesting that bothgroups have similar patterns of cognitive impairment. How-ever, data on the number of test scores below the 10th percentileand on the extent of the cognitive impairment suggests thatmany of the transfrontal patients are functioning at a low levelon a greater number of tests.

On some tests the nonsurgical patients perform in a similarway to the control group, although over 40% show evidence ofcognitive impairment. The difficulties that these patients haveappear to be restricted to tasks that require higher levels ofcognitive processing.

The fact that so many of the nonsurgical patients suffer frommild cognitive impairment suggests that not all the deficits canbe attributed to surgical intervention, raising the possibilitythat the primary disease or hormone abnormalities secondaryto the tumour and/or its treatment could be responsible.Alternatively, the cognitive impairment could be a result ofnonspecific psychological factors associated with having achronic illness.

Radiotherapy was not found to be associated with increasedrisk of cognitive impairment. In fact, the results suggested thereverse for transfrontal patients, with those who had not hadradiotherapy apparently having poorest test performance. Thisrather counterintuitive pattern of results may be explained bytwo factors, the effects of tumour regrowth or radical surgery.Patients treated with transfrontal surgery often receive thattreatment approach because their tumours are considered to belarge and therefore more vigorous than is the case for patients

Cognitive dysfunction in patients with pituitary tumour 395

q 1998 Blackwell Science Ltd,Clinical Endocrinology, 49, 391–396

Table 5 Percentage of patients with cognitiveimpairment Category Transfrontal Transsphenoidal Non-surgical Control

None 39 30 57 91 Chi¼ 36·75Mild impairment 4 44 17 9 P<0·01Moderate impairment 13 9 13 0Severe impairment 44 17 13 0

Table 6 Mean scores on tests with significantinteraction effects between radiotherapy andsurgery

Transfrontal Transsphenoidal

Test Radiotherapy Mean SD Mean SD P

List learning Yes 10·8 3·2 9·8 4·2 0·04Immediate recall No 5·8 9·8 10·0 2·9List learning Yes 9·3 4·7 9·3 4.1 0.04Delayed recall No 4·5 4·6 10·1 2·9Trail making Yes 29·4 12·7 28·4 5·4 0·02Part A No 56·3 40·3 32·2 20·0

who have transsphenoidal surgery. The purpose of radiotherapyis to prevent tumour re-growth where the surgeon believessome tumour tissue may remain. Hence it is possible that someof the patients who have not had radiotherapy may have sometumour re-growth, which is impinging on surroundingstructures and/or causing greater endocrine disturbance,which is in turn affecting cognitive functioning. An alternativehypothesis is that where more radical surgery is performed thesurgeon is confident that all tumour tissue has been removed,therefore making radiotherapy unnecessary, but having theside-effect of removing vital tissues necessary for efficientcognitive functioning. Similar findings were reported byThomsett et al. (1980) in childhood craniopharyngiomapatients. Of 42 surgery patients reviewed they found poorestoutcome in those who had not had radiotherapy, and concludedthat a combined surgical/radiotherapy approach should betaken.

Drawing unambiguous conclusions about the aetiology of thedifferences between the three treatment groups was not possiblefrom the present studies, however, and while differences appearto be related to different methods of treatment they may reflectdifferences that were present before treatment was carried out.There are known differences in tumour type and tumourcharacteristics which guide clinicians in their choice oftreatment. A prospective study which assesses psychologicalstatus from point of entry into the medical system throughto follow-up after treatment would assist in determiningthe relative contribution of the tumour and/or treatmentvariables.

Nevertheless, the present study provides useful informationabout the neuropsychological difficulties facing patientstreated for pituitary tumour, differences between thosereceiving different types of treatment, and reinforces the needfor adequate counseling for patients and family membersabout the likely psychosocial consequences of their illness.Although it is not possible to draw conclusions about thecause of the cognitive impairments on the basis of thepresent data, there is sufficient evidence available to rethinksome aspects of conventional clinical wisdom, and to set upsome testable hypotheses which will address causality.

Acknowledgements

This project was supported by Pharmacia and Upjohn

References

Aristizabel, S., Caldwell, W.L. & Arila, J. (1977) The relationship oftime–dose fractionation factors to complications in the treatment ofpituitary tumours by irradiation.International Journal of Radiation,Oncology and Biological Physics, 2, 667–673.

Army Individual Test Battery (1944)Manual of Directions and Scoring.War Department, Adjutant General’s Office, Washington DC.

Atkinson, A.B., Allen, I.V., Gordon, D.S., Hadden, D.R., Maguire,C.J.F., Trimble, E.R. & Lyon, A.R. (1979) Progressive visual failurein acromegaly following external pituitary irradiation.ClinicalEndocrinology, 10, 469–479.

Benton, A.L & Hamsher, K. de S. (1976)Multilingual AphasiaExamination.University of Iowa, Iowa City.

Bleehan, M. & Glastein, E. (1983)Radiation Therapy Planning, 1stedn. Marcel Dekker Incorporated, New York.

Flickinger, J., Nelson, P.B., Taylor, F.H. & Robinson, A. (1989)Incidence of cerebral infarction after radiotherapy for pituitaryadenoma.Cancer, 63, 2404–2408.

Grattan-Smith, P.J., Morris, J.G.L., Shores, E.A., Batchelor, J. &Sparks, R.S. (1992) Neuropsychological abnormalities in patientswith pituitary tumours.Acta Neurologica, 86, 626–631.

Horwitz, N.H. & Rizzoli, H.V. (1982) Intracranial neoplasm’s. InPostOperative Complications of Intracranial Neurological Surgery(edsN.H. Horwitz & H.V. Rizzoli). Williams and Wilkins, Baltimore.

Mawk, J.R., Ausman, J.I. & Erickson, D.L. (1979) Vasospasmfollowing transcranial removal of large pituitary adenomas. Reportof 3 cases.Journal of Neurosurgery, 50, 229.

Nelson, H.E. (1982) The National Adult Reading Test.Test Manual,NFER-Nelson, Windsor, UK.

Peace, K.A., Orme, S.M., Thompson, A.R., Padayatty, S., Ellis, A.W. &Belchetz, P.E. (1997) Cognitive dysfunction in patients treated forpituitary tumours. Journal of Clinical and Experimental Neuro-psychology, 19, 1–6.

Rey, A. (1964)L’examen Clinique en Psychologie.Presses Universi-taires de France, Paris.

Saville, D.J. (1990) Multiple comparison procedures: the practicalsolution.American Statistician, 44, 174–180.

Sheline, G.E. (1974) Treatment of non-functioning chromophobe adenomasof the pituitary.American Journal of Radiology, 120,553–561.

Sheline, G.E., Warra, W. & Smith, V. (1980) Therapeutic irradiationand brain injury.International Journal of Radiation Oncology andBiological Physics, 6, 1215–1228.

Thomsett, M.J., Conte, F.A., Kaplan, S.L. & Grumbach, M.M. (1980)Neurologic and endocrine outcome in childhood craniopharyngioma.Review of effect of treatment in 42 patients.Journal of Paediatrics,97, 728–735.

Warrington, E.K. (1984)Manual of the Recognition Memory Test.NFER-Nelson, Windsor, UK.

Wechsler, D. (1945) A standardised memory scale for clinical use.Journal of Psychology, 19, 87–95.

Wechsler, D. (1981)WAIS-R Manual. Psychological Corporation, NewYork.

396 K. A. Peace et al.

q 1998 Blackwell Science Ltd,Clinical Endocrinology, 49, 391–396