neurosurgery, psychiatry cerebral perfusion psychometric ... · 36ournalofneurology, neurosurgery,...

36ournal of Neurology, Neurosurgery, and Psychiatry 1995;59:368-374

Cerebral perfusion and psychometric testing inmilitary amateur boxers and controls

P M Kemp, A S Houston, M A Macleod, R J Pethybridge

AbstractThe objective was to compare two neuro-

physiological variables in active amateur

boxers with non-boxing sportsmen. 41boxers and 27 controls were given psy-

chometric tests: 34 boxers and 34 controlsunderwent technetium-99m hexamethyl-propyleneamineoxime single photonemission computerised tomography (Tc-99m HMPAO SPECT) cerebral perfusionscans. The controls performed better atmost aspects of the psychometric tests.Boxers who had fought fewer bouts had a

tendency to perform better at psychomet-ric tests than those boxers who hadfought more bouts. Tc-99m HMPAOSPECT cerebral perfusion scanningshowed that controls had less aberrationsin cerebral perfusion than the boxers.

In conclusion, significant differenceswere shown in two neurophysiologicalvariables between young amateur sports-men who box and those who do not. Thelong term effects of these findings remainunknown.

(7 Neurol Neurosurg Psychiatry 1 995;59:368-374)

Keywords: boxing; psychometrics; SPECT

Department ofNuclear Medicine,Royal Naval Hospital,Haslar, Gosport,HampshireP M KempA S HoustonM A MacleodDepartment ofStatistics, Institute ofNaval Medicine,Alverstoke, Gosport,HampshireR J PethybridgeCorrespondence to:Dr P M Kemp, Departmentof Nuclear Medicine, RoyalNaval Hospital, Haslar,Gosport, Hampshire.Received 8 March 1995and in revised form16 May 1995Accepted 1 June 1995

The "punch drunk syndrome"-dementiapugilistica embodies the chronic sequela ofboxing and has certain features in common

with Alzheimer's dementia and Parkinson'sdisease.' 3 For over 25 years it has beenknown that professional boxing is associatedwith this debilitating neurological disease';however, there is considerable controversy

about the long term effects of amateur boxing.Dementia pugilistica is characterised by disor-ders of movement-a consequence of damageto pyramidal, extrapyramidal, and cerebellarpathways-and the dementia is secondary to

widespread damage throughout the brain.Mechanisms of damage include contusion,shearing forces on nerves and vessels, and dif-fuse amyloid deposition of amyloid protein.Immunohistochemical staining of this amy-

loid has clearly shown the histological similar-ities of the punch drunk syndrome withAlzheimer's dementia.34The few studies in the scientific literature

that assess the safety of amateur boxing pro-duce conflicting results.5-l4 Extracting the datafrom mixed studies on amateur and profes-sional boxers generally shows adverse findings

in the amateur boxers,' 18 but numbers studiedwere too small (under 10) to derive substan-tive conclusions.A diversity of investigations have been

undertaken to detect subtle neurologicalimpairment in fit amateur boxers, includingCT, MRI, EEG, evoked potentials, and psy-chometric testing. This reflects a lack of con-sensus among investigators as to which is themost appropriate test to use, if any.Furthermore, most studies are compromisedto a varying degree by selection bias, absenceor lack of appropriate controls, the presenceof extraneous head injuries inflicted outsidethe ring, and multiple investigatory proce-dures that increase the possibility of chancefindings.

Amateur boxing is encouraged in thearmed forces to promote fitness and self disci-pline. To ascertain whether boxers represent-ing the Royal Navy and Army boxingassociations exhibit signs of subtle neurologi-cal abnormalities, this prospective study wasundertaken to compare a group of amateurboxers with appropriate servicemen controls.From the gamut of procedures available forinvestigating the effects of closed headinjury the potential insult encountered inboxing-it was decided to use psychometrictests and technetium-99m hexamethylpropy-leneamineoxime single photon emission com-puterised tomography (Tc-99m HMPAOSPECT) cerebral perfusion imaging. The Tc-99m HMPAO SPECT provides informationon regional cerebral perfusion, but by itselfdoes not provide an absolute measurement ofblood flow. The tomographic images pro-duced are akin to those obtained by PET,which determines cerebral blood flow with15Q. 120 As ischaemic pathophysiologicalchanges are known to occur after closed headinjury,2' 22 it is only to be expected that cere-bral perfusion imaging of the brain with Tc-99m HMPAO SPECT will be more sensitivein showing the early subtle abnormalities thanthe subsequent structural changes, if any, thatmay be detected by CT or MRIF2'2

MethodsSELECTION OF BOXERSAmateur boxers who represent either theRoyal Navy or Army boxing associations wereinvited to participate in this study. At the timeof request all the amateur boxers who repre-sented their respective services volunteeredand gave informed consent, thus minimisingselection bias. None of the boxers had sus-

368

on January 18, 2021 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.59.4.368 on 1 October 1995. D

ownloaded from

http://jnnp.bmj.com/

Cerebral perfusion and psychometric testing in military amateur boxers and controls

tained a relevant head injury outside the ring(defined as more than momentarily loss ofconsciousness or concussion lasting more thana minute) or had a history of neurological ill-ness. Most of the boxers underwent both psy-chometric tests and Tc-99m HMPAO SPECTbrain scanning; however, due to logistical con-straints a few boxers were not able to haveboth investigations. In total 42 boxers partici-pated in the trial. Forty one boxers underwentpsychometric tests of which 33 also had Tc-99m HMPAO SPECT scans; one boxerunderwent Tc-99m HMPAO SPECT withoutpsychometric testing. The boxers varied in agefrom 18 to 30 years, number of competitivebouts fought ranged from 0 to 250 (median40), and duration of participation in boxingwas from three months to 17 years.

SELECTION OF CONTROLSServicemen undergoing medical assistanttraining were invited to participate as controlsin the study. Criteria for selection includedage 18-30 years, regular participation in ama-teur sports, no history of relevant head injury(as defined for the boxers), no history of neu-rological illness, and no previous participationin boxing. In total, 34 controls underwent Tc-99m HMPAO scanning and a further 27 con-trols did the psychometric tests.

TC-99M HMPAO SPECT SCANNING TECHNIQUEAND INTERPRETATION OF IMAGEAll subjects had 550 MBq of Tc-99mHMPAO injected intravenously into an ante-cubital vein in a dimly lit room and werescanned within 10 minutes of injection with aSiemens 7500 ZLC tomographic camera witha low energy all purpose collimator linked to aBartec Micas V Sun System. Data wereacquired for 15 seconds from each of 64angles into a 140 keV + 10% window.Filtered back projection was performed with asoft Shepp-Logan filter and the data stored as64 axial slices of 64 x 64 matrices. Theimages produced of the boxers and controlswere compared with an "atlas of normality" todetect focal abnormalities. The atlas of nor-mality was derived from the images of 50healthy male and female subjects, civilian andmilitary personnel, who were aged 18-60years and had no history of head injury orneurological illness. These 50 images wereregistered with respect to a reference image25and normalised to the cerebellar counts. Thenormal atlas was created with a mean and sev-eral principal component images after regis-tration and normalisation.26 The resolution ofthe mean and component images was main-tained at the voxel level of the original images.The principal component images correspondto the most significant patterns of variation inthe 50 images, and it was found that six com-ponents were necessary to provide an ade-quate description of normal variation. Theregistered normalised images of the 34 boxersand 34 controls were then compared auto-matically to the atlas of normality.2627 Thesignificance (in normal SDs) of voxels withreduced perfusion in the cortical region was

calculated. An image was defined as abnormalif it contained at least one deficit with 10 con-nected voxels above a threshold of 2-9 SDs.

PSYCHOMETRIC TESTSThe battery of tests selected assessed a widerange of cognitive functions28 (reaction time,psychomotor learning, reasoning, perceptionand comparison, short term memory, andmotor function). The test battery included thefollowing tasks:

Four choice reaction timeFour boxes numbered 1-4 are displayed onthe screen and one of the boxes is highlighted.The participant has to strike the appropriatenumerical key. The mean response latency forcorrect answers was recorded ( the greater thetime the slower the response).

Code substitutionThe code consists of nine letters and corre-sponding numbers displayed across the top ofthe screen. Underneath the code rows of let-ters are displayed and the participant has toinsert the appropriate number. The numberof letters correctly encoded was recorded.

Logical reasoningLogical statements are presented on thescreen and the participant has to decidewhether these statements are true or false.Statements can be expressed in the affirmativeor negative. Verbs include "trails", "follows",and "precedes". The percentage of statementscorrectly answered was recorded.

Simultaneous pattern comparisonPairs of patterns are displayed simultaneouslyon the screen; the participant has to decidewhether these are the same or different. Thenumber of correct answers was recorded.

SternbergFour numbers are displayed on the screen forone second. Subsequently, pairs of numbersare displayed and the participant has to decidewhether at least one of the pair of numberswas represented in the original set of fournumbers. The mean response latency time forcorrect answers was recorded-that is, thegreater the time the slower the response.

Finger tappingThe participant has to press two adjacent keyswith the index and middle finger of the non-dominant hand. The number of alternatepresses was recorded.

The automated test battery was given by anexperienced person. All tasks were assessedover a two minute period except for the alter-nate finger tapping, which lasted for 30 sec-onds. Forty one boxers and 27 controls hadthree practice attempts at each of the testsbefore the main run. A Zenith lap top com-puter was used throughout the study and alltests were undertaken in a quiet, well lit, andwarm room. Ability at some aspects of psycho-metric testing can be influenced by age and a

369 on January 18, 2021 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.59.4.368 on 1 O

ctober 1995. Dow

nloaded from


Kemp, Houston, Macleod, Pethybridge

subject's intellect;29 consequently, it may benecessary to adjust for these factors. It wasdecided to use school attainment in the formof number of 0 level examinations passed as amarker of intellect. Statistical modelling wouldshow whether age and intellect would be suit-able as covariates in the analysis.

This study received approval from theRoyal Naval clinical research scientific com-mittee (military), the local ethics committee(civilian), and the administration of radioac-tive substance advisory committee. All volun-teers gave informed written consent.

ResultsAGE AND SCHOOL ATTAINMENTTable 1 shows the summary of the data ofthe distributions of the ages and number of0 levels for the amateur boxing and controlgroups. Applying the Kolmogorov-Smirnovtest to these cumulative distributions for thegroups of boxers and controls showed nosignificant differences (Kolmogorov-Smirnovtest P > 0 1).

TC-99M HMPAO SCANNINGInitial data exploration (Zelen test) showedthat the scan findings (normal or abnormal)

Table 1 Summary data of the distributions of the agesand number of0 levels for the amateur boxing and controlgroups undergoing Tc-99m HMPAO SPECT andpsychometric testing

No Mean (SD) Minimum Maximum

Tc-99m HMPAO SPECTAge:

Boxers 34 22 0 (2 9) 18 29Controls 34 22 2 (3 1) 18 29

O levels:Boxers 34 1 9 (2 6) 0 8Controls 34 3 1 (2 8) 0 9

Psychometric testingAge:

Boxers 41 22-0 (2 8) 18 29Controls 27 22.0 (2 9) 18 29

O levels:Boxers 41 2-4 (3 1) 0 8Controls 27 3-1 (2 6) 0 9

Table 2 Results of comparing HMPAO SPECT ofamateur boxers and controls with the "atlas of normality"

HMPAO SPECT

Normal Abnormal Total

Boxers 20 14 34Controls 29 5 34

Table 3 Average psychometric test scores for 41 boxers and 27 controls withoutadjustment for covariates

(95% CIfor difference ofPsychometric test Group Mean means (boxers - controls)

Four choice reaction time 41 boxers 0 49 s (0-04 to 0-10)*27 controls 0-42 s

Code substitution 41 boxers 46-5 correct (-83 to-1-1)*27 controls 51-2 correct

Pattern recognition 41 boxers 89-9 correct (-17 6 to -2 0)*27 controls 99-7 correct

Steinberg reaction time 41 boxers 1-18 s (0 0 to 0-2)27 controls 1 11 s

Alternate finger tapping 41 boxers 90 taps (- 33 to - 13)*27 controls 1 13 taps

*p < 005.

were not influenced by age or school attain-ment. Table 2 shows the summary of theresults of Tc-99m HMPAO SPECT scanningfor the groups of boxers and controls.Abnormality is defined as the presence of oneor more regions of significant aberration incerebral perfusion when compared with theatlas of normality. It can be seen that 41% ofthe amateur boxers had abnormal scans, com-pared with 14% of the controls (Z = 5.9,df = 1, P < 0 02). In the control group theaberrations in cerebral perfusion were largelysingular whereas in the boxing group the aber-rations were generally multiple. In bothgroups there were no predominant sites ofabnormality, the lesions were scatteredthroughout the cerebral cortex.

PSYCHOMETRIC TESTINGTable 3 shows the average psychometric testscores of the raw data (before transformationor introduction of covariates) for the groupsof boxers and controls. The logical reasoningtask had to be excluded from the analysis asmany boxers and several controls had scoresaround 50%. This suggested that those sub-jects found the task too difficult and wereguessing. The distributions of the scores forthe code substitution, pattern recognition,and finger tapping tasks for each group ofboxers and controls were found to be normalby univariate and multivariate analysis. Theresults of the reaction time and Stemnbergtasks required reciprocal transformations tosatisfy normality assumptions in the statisticalmodels considered.

UNIVARIATE ANALYSIS OF VARIANCE (ANOVA)The results of a univariate ANOVA show thatin all psychometric tests analysed the boxersperformed less well, on average, than the con-trols. In four of the five tests these differencesranged from 10-20% and were significant(P < 0.05).

UNIVARIATE ANALYSIS OF COVARIANCE(ANCOVA)Linear modelling testing of the data (or trans-formed data) for each psychometric test withmain effects for age, age squared, number ofO levels and group (boxer or control), andinteractions between group and education,group and age, and group and age2, indicatedthat the covariates (age, age2, and number ofO levels) should be included in the model.Also, the interaction terms could be excludedas they had no significant influence on thepsychometric tests.

Table 4 shows a summary of the averagepsychometric test scores (untransformed)with the adjustment for the covariates. In fourout of the five tasks there were significant dif-ferences between the groups to the detrimentof the boxers. These differences ranged from10-20%.

COMPARISON OF BOXERS WITH A LOW ANDHIGH NUMBER OF BOUTSThe median number of bouts fought can beused to divide the boxers into two groups; 21



j.com/

J Neurol N


ctober 1995. Dow

nloaded from



Table 4 Summary of average psychometric test scores for 41 boxers and 27 controlsadjustedfor covariates (age, age2, number of0 levels)

(95% CIfor difference ofPsychometric test Group Mean means (boxers - controls)

Four choice reaction time 41 boxers 0 49 s (0-04 to 0 10)*27 controls 0 42 s

Code substitution 41 boxers 46 6 correct (-8-0 to -09)*27 controls 51-0 correct

Pattern recognition 41 boxers 90-2 correct (- 15 7 to -2-1)*27 controls 99-1 correct

Stemnberg reaction time 41 boxers 1-17 s (-01 to 0 2)27 controls 1-12 s

Alternate finger tapping 41 boxers 90 taps (-32 to - 12)*27 controls 113 taps

*p < 0 05.

Table 5 Summary of average psychometric test scores for 21 low bout boxers and 20 highbout boxers adjustedfor covariates (age, age', number of 0 levels)

(95% CIfor difference ofPsychometric test Group Mean means (low - high bouts)

Four choice reaction time 21 low bout 0-46 s (-0 10 to -001)*20 high bout 0-52 s

Code substitution 21 low bout 48-3 correct (-1-3 to -87)20 high bout 44-6 correct

Pattern recognition 21 low bout 94-2 correct (0 7 to 17-1)*20 high bout 85-3 correct

Alternate finger tapping 21 low bout 88 taps (-16-0 to 8 6)20 high bout 92 taps

*P < 005.

boxers with a maximum of 40 bouts (lowbouts) and 20 boxers with more than 40 bouts(high bouts). Table 5 summarises averagepsychometric test scores (untransformed).Excluding the Sternberg test due to an inter-action between groups (low and high bouts)and education, a univariate ANCOVAshowed that low bout boxers performedsignificantly better than the high bout boxerson the four choice reaction time and patternrecognition (P < 0 05). There were no

significant differences on the remaining twotasks.

Treating the number of bouts as a continu-ous variable showed a significant inverse rela-tion between the four choice reaction time(reciprocal values) and the number of boutswhen regressed in the presence, or absence, ofthe covariates (P < 0 05). Similarly, therewas a significant inverse relation between pat-tern recognition and number of bouts(P < 005).

MULTIVARIATE ANALYSIS OF COVARIANCEMultivariate ANCOVA with the psychometrictest scores (transformed where appropriate) asvariates and taking age, age2, and number ofO levels as covariates, showed a significantdifference between the control and boxergroups (P < 0 05). Multivariate techniquesalso disclosed significant differences betweenthe low bout boxers and the control group(P < 0 05) and a tendency for a differencebetween the boxers with low and high bouts(P < 0.1).

DiscussionAnimal studies have shown that multiple sub-concussive blows to the head cause greaterneurological and vascular damage than lessfrequent knock out blows.30 Postmortemexamination of the brains of the patients who

had endured minimal head trauma but subse-quently died from other than cranial causeshave shown microglial scars.3" Also, psycho-metric testing has been shown to elicit thedetrimental effects of cumulative concus-sion.32 Consequently, it is not surprising thatsome think that the term "commotiocerebri"-brief loss of consciousness withoutpermanent brain damage-should be dis-carded.2'

Given the ischaemic pathophysiologicalchanges found in closed head injury2 22 it isnot surprising that functional brain imagingwith Tc-99m HMPAO SPECT scanning ismore sensitive than CT or MRI at showingthe consequences of closed head injury.2324Cerebral amyloid protein deposition,immunohistochemically similar to that seen inAlzheimer's disease, has been noted in thebrains of ex-boxers3 4 and in people with headtrauma.33 Most importantly, it has also beenshown that these regions of amyloid proteindeposition are associated with abnormal vas-culature,34 which could also account for theabnormalities noted on Tc-99m HMPAOSPECT cerebral perfusion imaging.The development of a computer program

to detect aberrations in Tc-99m HMPAOSPECT cerebral perfusion imaging removesthe potential for human bias. This programuses a multivariate technique to assess cere-bral perfusion. Similar techniques have beenapplied to the interpretation of PET scansthat show cerebral metabolism.35-37 Computeranalysis of the images in this study showedfive of the 34 controls to have aberrations incerebral perfusion patterns giving a specificityof 85%. This value compares favourably withthe specificity of 84% noted by PET scan-ning.38 Comparing the cerebral perfusion pat-terns of the amateur boxers with the controlsin our study showed a significantly greaterproportion of aberrations in the cerebral per-fusion patterns of boxers than in the controls.

Although there was not a significant differ-ence between the psychometric test scores ofthe 13 boxers with abnormal cerebral perfu-sion and the 20 boxers with normal scans(P = 0-2) this was not unexpected. Firstly,the numbers were small, and secondly,HMPAO cerebral perfusion imaging and psy-chometric testing are assessing differentaspects of neurophysiology. It is only to beexpected that, in the detection of subtle neu-rological abnormalities, focal cerebral perfu-sion deficits may occur in areas of the cerebralcortex that are not required for the cognitivetasks of psychometric testing. In overt neuro-logical disease-for example, patients withAlzheimer's disease or severe head injury-theresults ofHMPAO cerebral perfusion imagingand psychometric testing (and other neuro-logical investigations) are more likely to showconcordance.3941 It is arguable whether thesame controls that underwent psychometrictesting should also have had the HMPAObrain scans. At the outset of this trial the mainconsideration was to compare two groups ofsportsmen (boxers and non-boxers).Consequently, it was logistically easier to

371


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from



recruit separate control groups. Given the rel-atively high specificity ofHMPAO brain scan-

ning, 85% in this study, it would haverequired considerably more controls (and eth-ical deliberations) to undertake a study of suf-ficient power to compare HMPAO cerebralperfusion imaging with psychometric testingin young healthy adults.

Psychometric testing is a highly sensitiveinvestigation for detecting the subtle cognitiveimpairment after head injury.3242 When usingpsychometric testing in a group comparativestudy it would seem appropriate to recruitcontrols of a similar age, intellectual ability,and possibly other social factors. In this studywe recruited controls who were sports playingservicemen who were of a similar age and hadattained similar educational qualifications.Intelligence has been shown to correlate mod-erately with some psychometric tests29 (thecorrelation coefficient being of the order of0 5). A similarly significant correlation hasbeen shown between school attainment andintelligence;43 thus it was decided to use

school attainment as a potential marker ofintellect. Obviously, in any group comparativestudy it is preferable to match for importantvariables beforehand rather than control sta-

tistically for differences subsequently. Astraight forward comparison of the psychome-tric test scores of the groups of boxers andcontrols, without the introduction of covari-ates, showed significant differences betweenthe groups in four of the five psychometrictests to the detriment of the boxers. Statisticalmodelling of the psychometric test scores

showed that the covariates of age, age2, and

school attainment should be introduced in themodel. Statistical analysis with these covari-ates included showed almost identical resultsto the analysis without covariates-namely, infour out of five tests the boxers performedsignificantly worse. The similarity of theresults of these different types of analyses(with and without covariates) was not unex-

pected given the similar distributions of ages

and school attainment of the two groups.Furthermore, although on average the con-

trols had achieved slightly, but not

significantly, better at school, the introductionof the covariates would have adjusted the box-ers psychometric test results favourably rela-tive to those of the controls. It cannot becompletely excluded that the introduction of a

more formal measure of intellect (as definedby detailed intelligence assessment) as a fur-ther covariate, as well as age, age2, and num-

ber of 0 levels, would have accounted forsome of the residual variance in the model.Similarly, it could be argued that other covari-ates should have been assessed-for example,social class of participant's mother (this couldreflect educational opportunity) or even

amount of time participating in computergames (a task not too dissimilar to automatedpsychometric testing). At the outset of thistrial it was the intention of the investigators to

minimise the time required by the boxers andcontrols to undergo the investigations. Thereis no doubt that the additional introduction of

formal evaluation of intelligence would havereduced the numbers who volunteered withthe consequent introduction of a selectionbias. Unfortunately, in all scientific studiesthat require volunteers, there will always haveto be a trade off between the components ofthe potential biases in the methodology.

Butler et al'3 have recently published thefindings of a United Kingdom study compar-ing the psychometric test scores of 86 amateurboxers with those of 31 water polo players and47 rugby union players. Comparing theresults of the boxers with those of the controlswithout the introduction of any covariatesshowed that the boxers performed worse onall 12 psychometric tests; in eight of thesetests the differences were significant. Theauthors noted that many of the boxers wereearly school leavers whereas a proportion ofthe rugby and water polo players were under-graduates; consequently, a definitive analysiscould not be undertaken as no estimates wereavailable for intellect. In the absence of beingable to analyse the psychometric data, andgiven that other neurological investigationsgave negative results, the investigators con-cluded that "no evidence of neuropsychologi-cal dysfunction due to boxing could befound". Surely it would have been moreappropriate to have selected controls of a sim-ilar scholastic level rather than attempt toadjust the psychometric test scores statistically(even if appropriate data had been collected)for what seems to be two groups of widely dif-fering intellectual ability?

In our study the number of bouts correlatedhighly with the number of years of boxing.Consequently either variable could be used as ameasure of boxing exposure. The group ofboxers could be divided around the mediannumber of bouts into those with less than andthose with more than 40 bouts. Univariate andmultivariate analyses of covariance showed dif-ferences in cognitive performance between thetwo groups to the detriment of those boxers inthe over 40 bout group. These findings couldalso be shown by significant inverse trends incertain aspects of psychometric testing andnumber of bouts fought. Similar findings werenoted by Stewart et al. 14 This prospectiveAmerican study conducted between 1986 and1990 spent $1-5 million in examining 484amateur boxers over a four year period.Statistical modelling of the psychometric datain the American study showed that age, age2,and education, together with other covariatescould be used in the model. It is interestingthat common covariates were noted in this andour study. Psychometric testing of the 484amateur boxers on entry to the study showedsignificant tests of trend between the totalnumber of bouts fought and ability in psycho-metric testing. Although follow up testing on393 of the boxers over a two year period didnot show any deterioration in psychometricperformance, the investigators urged cautionin the interpretation of this finding as a longerlatency period may be required before effectsare manifested. Similarly, Butler et al'3 did notshow any deterioration in their boxers on

372


http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from



Table 6 Major studies assessing the safety of amateur boxing

Investigators Participants Findings Comments

Thomassen et al5 53 fQrmer boxers No significant differences on clinical These findings are confounded by theDENMARK 1979 53 footballers examination, EEG, or psychometric fact that 88% of the controls and 75%

testing. of the boxers sustained head injuriesoutside the ring.

McLatchie et at6 20 active boxers Abnormal clinical examination in 35% No controls had clinical examinationSCOTLAND 1987 Variable number of boxers, abnormal EEG in 40% of or EEG. University staff used as

of controls boxers, abnormal psychometric testing controls for psychometric testing.in 63% of boxers.

Brooks et al7 29 active boxers No significant differences on Only 39% of invited boxersSCOTLAND 1987 19 controls psychometric testing. participated. Controls matched for age,

education, and ethnicity.Haglund et al9 12 50 former boxers No significant differences on clinical Although the authors concluded thatSWEDEN 1990 50 athletes examination, CT, MRI. EEG showed amateur boxing appeared "safe" the

significantly greater abnormalities in differences on EEG examinationboxers. remain unexplained.

Butler et all' 86 active boxers Significant differences between boxers No allowance for the intellectualENGLAND 1993 31 water polo and and controls on initial psychometric differences between the controls (some

47 rugby players testing to the detriment of the boxers. were university students) and theImprovement noted on longitudinal boxers (many early school leavers).assessment. No differences on clinical Authors concluded no evidence ofexamination, evoked responses, EEG, damage in boxers.CKBB isoenzymes, or ophthalmologicalassessment.

Stewart et al"4 484 amateur boxers Significant tests of trend between Authors urge caution in theUSA 1994 psychometric testing ability and interpretation of the longitudinal

number of bouts. No deterioration on findings as a longer latency period may2 year follow up. be required before effects are manifest.

Kemp et al (i) 34 active boxers Boxers had significantly more All invitees participated thus reducingENGLAND 1994 34 controls abnormalities on Tc-99m HMPAO selection bias. Subjects with(This study) (ii) 41 active boxers cerebral perfusion imaging. extraneous head injuries were not

27 controls Psychometric testing revealed boxers included. Controls matched for ageto perform significantly worse than and education.controls and, in addition, those boxerswith the greater number of boutsperformed significantly worse thanthose with few bouts.

longitudinal assessment over an undisclosedperiod. They did note that "any effects due toboxing would have had to be large to bedetected against the background of premorbidvariation in ability". Unfortunately, they didnot have any data available for "estimates ofintellectual capacity".

Table 6 lists the results of recent majorstudies that assessed the safety of amateurboxing. It can be seen that our study has beenrelatively more fortunate than most in termsof selection bias, extraneous head injuries,and choice of controls.

In summary, psychometric testing and Tc-99m HMPAO cerebral perfusion imagingwere undertaken in a group of service amateurboxers and a control group of sportsmen.Psychometric testing showed detrimentalfindings in the group of boxers and this wasmore apparent in those who had the mostbouts. Computer analysis of Tc-99mHMPAO SPECT images showed significantlygreater aberrations in cerebral perfusion in thegroup of boxers than the controls. This studyhas noted subtle impairment in two aspects ofneurophysiology in those young men whoreceive recurrent head trauma from amateurboxing. The question remains, and is of theutmost importance, as to whether thesefindings are reversible or will have an effect onsubsequent mental, physical, or social wellbe-ing. It is hoped that the investigators of the$1 5 million study in the United States'4 willfollow up their large cohort of boxers to pro-vide a definitive answer.

1 Roberts AH. Brain damage in boxers. London: Pitman,1969.

2 Corsellis JAN, Bruton CJ, Freeman-Browne D. The after-math of boxing. PhysiolMed 1973;3:270-303.

3 Roberts GW, Allsop D, Bruton C. The Occult aftermathof boxing. J Neurol Neurosurg Psychiatry 1990;53:373-8.

4 Tokuda T, Ikeda S, Yanagisawa N. Re-examination of ex-

boxers' brains using immunohistochemistry with anti-bodies to amyloid B protein and tau protein. ActaNeuropathol 1991;82:280-5.

5 Thomassen A, Juul-Jenson P, Olivarius B, Braemer J,Christensen A. Neurological, electroencephalographicand neurophysiological examination of 53 former ama-teur boxers. Acta Neurol Scand 1979;60:352-62.

6 McLatchie G, Brooks N, Galbraith J, et al. Clinical neuro-logical examination, neuropsychology, electroen-cephalography and computerised tomographic headscanning in active amateur boxers. J Neurol NeurosurgPsychiatry 1987;50:96-9.

7 Brooks N, Kupshik G, Wilson I, Galbraith S, Ward R. Aneuropsychological study of amateur boxers. J NeurolNeurosurg Psychiatry 1987;50:997-1000.

8 Jordan BD, Zimmerman RD. Magnetic reasonance imagingin amateur boxers. Arch Neurol 1988;45:1207-8.

9 Haglund Y, Edman G, Murelius 0, et al. Does Swedishamateur boxing lead to chronic brain damage? 1. A ret-rospective medical, neurological, and personality traitstudy. Acta Neurol Scand 1990;82:245-52.

10 Haglund Y, Bergstrand G. Does Swedish amateur boxinglead to chronic brain damage? 2. A retrospective studywith CT and MRI. Acta Neurol Scand 1990;82;297-302.

11 Haglund Y, Pearson HE. Does Swedish amateur boxinglead to chronic brain damage? 3. A retrospective clinicalneurophysiological study. Acta Neurol Scand1990;82:353-60.

12 Murelius 0, Haglund Y. Does Swedish amateur boxinglead to chronic brain damage? 4. A retrospective neuro-physiological study. Acta Neurol Scand 1991;83:9-13.

13 Butler RJ, Forsythe WI, Beverly DW, Adams LM. Aprospective controlled investigation of the cognitiveeffects of amateur boxing. J Neurol Neurosurg Psychiatry1993;56: 1055-61.

14 Stewart WF, Gordon B, Selnes 0, et al. Prospective study ofcentral nervous system function in amateur boxers in theUnited States. Am J Epidemiol 1994;139:573-88.

15 Kaste M, Vilkki J, Sainio K. Is chronic brain damage inboxing a hazard of the past. Lancet 1982;ii:1 186-8.

16 Ross RJ, Cole M, Thompson JS, Kim KH. Boxers-com-puterised tomography, EEG, and neurological evalua-tion. JAMA 1983;249:211-3.

17 Casson IR, Siegel 0, Shan R. Brain damage in modemboxers. JAMA 1984;251:2663-7.

18 Jordan BD, Zimmerman RD. Computerised tomographyand magnetic reasonance imaging comparisons in box-ers. JAMA 1990;263:1670-4.

19 Gemmell HG, Evans NTS, Besson JAO, et al. Regionalcerebral blood flow imaging: a quantitative comparisonoftechnetium-99m HMPAO SPECT with C'502 PET. YNuclMed 1990;31:1595-1600.

20 Messa C, Perani D, Lucignani G, Zenorini A, Zito F,Rizzo G. High resolution technetium-99m-HMPAOSPECT in patients with probable Alzheimer's disease:comparison with fluorine-18-FDG PET. J Nucl Med1994;35:210-6.

21 van Zomeren AH. Reaction time and attention after closedhead injury. The Netherlands: Swets and ZeitlingerB V -Lisse, 1981.

22 Teasdale G, Mendelow D. Pathophysiology of head



j.com/

J Neurol N


ctober 1995. Dow

nloaded from



injuries. In: Brooks N, ed. Closed head injury.Psychological, social, and family consequences. Oxford:Oxford University Press, 1986.

23 Newton MR, Greenwood RJ, Britton KE. A study com-paring SPECT with CT and MRI after closed headinjury. Neurol Neurosurg Psychiatry 1992;55:92-4.

24 Roper SN, Mena I, King, et al. An analysis of cerebralblood flow in acute closed head injury using technetium-99m HMPAO SPECT and computed tomography. _J

Nucl Med 199 1;32:1684-7.25 Barber DC. Registration of low resolution medical images.

Phys Med Biol 1992;37:1485-98.26 Houston AS, Barber DC. Registration of a SPECT brain

image with its "nearest normal equivalent" image.Proceedings of the 4th international conference onimage processing and its applications. IEEE Conf Publ1992;354:369-72.

27 Houston AS, Kemp PM, Macleod M. A method forassessing the significance of abnormalities in HMPAObrain SPECT images. Jf Nucl Med 1994;35:239-44.

28 Smith M, ed. Users manual for the Essex Corporation auto-mated performnance test system version 5.03. Orlando,Florida: Essex Corporation, 1988.

29 Tumage JJ, Kennedy RS, Smith MG, Baltzley DR, LaneNE. Development of microcomputer-based mental acuitytests. Ergonomics 1992;35: 1271-95.

30 Unterharscheidt FJ. Injuries due to boxing and othersports. In: Vinken PJ, Bruyn GW, eds. Handbook of clini-cal neurology. Amsterdam: North Holland,1975;23:527-93.

31 Oppenheimer DR. Microscopic lesions in the brain follow-ing head injury. J7 Neurol Neurosurg Psychiatry1968;31 :299-306.

32 Gronwall D, Wrightson P. Cumulative effects of concus-sion. Lancet 1975;ii:995-7.

33 Gentleman SM, Graham DI, Roberts GW. Molecularpathology of head trauma: altered BAPP metabolismand the aetiology of Alzheimer's disease. Prog Brain Res1993;96:237-46.

34 Buee L, Hof PR, Bouras C, et al. Pathological alterations

of the cerebral microvasculature in Alzheimer's diseaseand related dementing disorders. Acta Neuropathol1994;87:469-80.

35 Clark C, Carson R, Kessler R, et al. Alternative statisticalmodels for the examination of clinical positron emissiontomography/flurodeoxyglucose data. J7 Cereb Blood FlowMetab 1985;5:142-50.

36 Clark CM, Ammann W, Martin WRW, et al. TheFDG/PET methodology for early detection of diseaseonset: a statistical model. .7 Cereb Blood Flow Metab1991;11:A96-A102.

37 Azari NP, Pettigrew KD, Schapiro MB, et al. Early detec-tion of Alzheimer's disease: a statistical approach usingpositron emission tomography data. J7 Cereb Blood FlowMetab 1993;13:438-47.

38 Fazekas F, Alavi A, Chawluk JB, et al. Comparison of CT,MR, PET in Alzheimer's dementia and normal ageing..7Nucl Med 1989;30: 1607-15.

39 Wiedmann KD, Wilson JTL, Wyper D, et al. SPECT cere-bral blood flow, MR imaging, and neuropsychologicalfindings in traumatic head injury. Neuropsychiatry1989;3:267-8 1.

40 Montaldi D, Brooks DN, McColl JH, et al. Measurement ofregional cerebral blood flow and cognitive performancein Alzheimer's disease. Neuro Neurosurg Psychiat1990;53:33-8.

41 McKeith LG, Bartholomew PH, Irvine EM, Cook J,Adams R, et al. Single photon computerised tomographyin elderly patients with Alzheimer's disease and multi-infarct dementia-regional uptake of technetiumlabelled HMPAO related to clinical measurements. Br_7Psychiatry 1993;163:597-603.

42 Levin HS, Amparo E, Eisenberg HM, et al. Magnetic rea-sonance imaging and computerised tomography in rela-tion to the neurobehavioural sequelae of mild andmoderate head injuries. JNeurosurg 1987;66:706-13.

43 Ainsworth ME. The relationship between motivation, person-ality, intelligence and school attainment in a secondarymodern school [thesis]. Manchester: University ofManchester, 1965.



j.com/

J Neurol N


ctober 1995. Dow

nloaded from


neurosurgery, psychiatry cerebral perfusion psychometric ... · 36ournalofneurology, neurosurgery,...

Documents