N. Horlock, G. D. Wilson, E M. Daley, R I. Richman, A. O Grobbelaar, R. Sanders and C. Foy

Transcription

1 British Jurnal f Plastic Surgery (1998), 51, The British Assciatin f Plastic Surgens BRITISH JOURNAL OF ~ PLASTIC SURGERY Cellular prliferatin characteristics d nt accunt fr the behaviur f hrrifying basal cell carcinma. A cmparisn f the grwth fractin f hrrifying and nn hrrifying tumurs N. Hrlck, G. D. Wilsn, E M. Daley, R I. Richman, A. O Grbbelaar, R. Sanders and C. Fy RAFT Institute f Plastic Surgery and The Gray Labratry Cancer Research Trust, Munt Vernn Hspital, Nrthwd, Middlesex, UK SUMMARY. This study cmpared the clinical features, histlgical subtype, grwth fractin (by Ki67 immunhistchemistry) and prliferatin pattern f 22 clinically defined hrrifying basal cell carcinma cmpared t 81 nn hrrifying lesins. Late presentatin was assciated with half f the hrrifying tumurs. The ther half develped hrrifying tumurs despite early interventin. The hrrifying tumurs exhibited a variety f histlgical grwth patterns. A ttal f 50% were infiltrative, 23% ndular and 18% micrndular. There was n difference in the grwth fractin r prliferatin pattern between hrrifying and nn hrrifying tumurs f similar grwth pattern (P = ns), althugh infiltrative tumurs in either grup exhibited a significantly higher grwth fractin than ndular tumurs (P < 0.01). This suggests that there is n intrinsic bilgical difference between hrrifying and nn hrrifying tumurs t accunt fr their behaviur. We cnclude that late presentatin, failed r inadequate early management especially f infiltrative turnuts (and ther subtypes) determines the develpment f hrrifying tumurs. Basal cell carcinma is defined by the Wrld Health Organizatin Cmmittee n the histlgical typing f skin turnuts as a 'lcally invasive slwly spreading tumur which rarely metastasises arising frm the epidermis r hair fllicle." The majrity f tumurs clinically appear t be slwly grwing and behave in a relatively benign, nn-aggressive fashin. 2 The recurrence rate is apprximately 10% but varies with treatment methd and recurrence rates may reach 40%? Rarely, sme turnuts d nt adhere t these general characteristics, becme increasingly destructive and recurrent and have been termed aggressive, 24 mutilating~ 7, giants.~ r hrrifying tumurs. ~~ Of these terms, hrrifying is defined the mst clearly as being large (greater than 3 cm), destructive (invading vital structures beynd the surrunding skin and subcutaneus fat eg cartilage, bne r brain), lcally uncntrllable (recurrence despite fur r mre adequate attempts at eradicatin) r metastatic. At present it is unknwn why sme tumurs behave in this fashin while thers fllw an apparently benign curse. Many factrs have been cnsidered imprtant in the evlutin f a hrrifying tumur: First, patient factrs such as age", race, '2 time f presentatin '3, hst immunity 14,j~ and the degree f expsure t envirnmental r ther carcingens. ~6 Secnd, tumur factrs such as the intrinsic tumur aggressiveness determined by the bilgy f each individual. Third, treatment factrs such as mdality '3 (raditherapy r surgery), adequacy f treatment r the accuracy f the histlgical reprt in terms f the margins f tumur excisin. N single r universal factr r cmbinatin f factrs have been causally related t hrrifying lesins. Cellular prliferatin characteristics are a fundamental bilgical determinant f the behaviur f many tumur types. They are f high prgnstic significance in breast cancer, varian cancers, neurblastma, bladder and lung cancers althugh f less prgnstic significance in clrectal cancer.'7 Tumur grwth is determined by a cmbinatin f the cell cycle time fr each tumur cell, the number f cells in the cell cycle at any particular time (the grwth fractin) and the balance f cell prductin with cell lss (eg frm necrsis, apptsis and desquamatin). While it is nt pssible t measure the cell cycle time in archival material, the grwth fractin can be measured by Ki-67 immunhistchemistry fllwing antigen retrieval techniques. '8 The Ki-67 antigen is a nn histne nuclear antigen present in G1, S and G2 phases f the cell cycle but nt in GO and can be measured using the MIB-1 mnclnal antibdy. In many turnuts the grwth fractin has been shwn t crrelate well with tumur behaviur eg breast, ~9'2~ lymphma, 2~,22 esphagus 23 and melanma? 4 There have been few studies investigating cell prliferatin in BCC Tritiated thymidine studies have indicated that there is greater prliferative activity in BCC than wuld be apparent frm the slw clinical grwth and that this is lcated at the periphery f the tumurs. Only three studies have used Ki 67 antigen as a marker f prliferatin in basal cell carcinma 25'263~ and nne f these has attempted t relate cell prliferatin t the hrrifying frm f this tumur. The Ki67 grwth fractin has been demnstrated t be higher in recurrent BCC and als in thse turnuts with infiltrative r mrpheic grwth pattern. 35 These histlgical 59

2 60 British Jurnal f Plastic Surgery subtypes are knwn t behave in an aggressive fashin. The aim f this study was t test the pstulate that hrrifying turnuts may have a higher grwth fractin than nn hrrifying tumurs and that this may be respnsible fr their mre aggressive clinical behaviur. Ptentially prblem lesins culd then be identified at an early stage. Ndular Micrndular Infiltrative Materials and methds Specimens Hrrifying tumurs. Jacksn's criteria were mdified since 3 cm was cnsidered t small and wuld nt exclude many superficial basal cell carcinmas. We defined hrrifying turnuts as: 1. Large, greater than 5 cm. All turnuts had t meet this criteria and include ne f the fllwing: 2. Destructive invading vital structures beynd the surrunding skin and subcutaneus fat, e.g. cartilage, bne r brain 3. Lcally uncntrllable recurrence despite fur r mre adequate attempts at eradicatin 4. Metastatic. Grwth f turnut in tissues nt in direct cntinuity with the primary site. Hrrifying cases were identified by persnal recllectin frm cnsultants based at the reginal plastic surgical unit at Munt Vernn Hspital. Sme f these patients had previusly been managed ver lng perids f time with multiple recurrences at varius ther hspitals and had prgressively develped hrrifying tumurs. An attempt was made t btain all f the specimens frm each recurrence s that the prgressin f these tumurs culd be studied. Nn hrrifying tumurs. This grup cnsisted f a cnsecutive series f patients, frm 1991, presenting at the reginal plastic surgical unit at Munt Vernn Hspital with primary BCC (nt recurrent tumurs). Patients with turnuts that were knwn t have recurred in the fllwing 5 years were excluded. In rder that histlgically similar turnuts in the hrrifying and nn hrrifying grups culd be cmpared, the rarer histlgical types (infiltrative, mrpheic and micrndular) f the nn hrrifying grup were augmented with BCCs frm Due t the rarity f hrrifying BCC, we were unable t increase the sample size f this grup. Histpathlgical classificatin J" tumur type In this study we have used a histpathlgical classificatin cmbining bth an assessment f the histlgical grwth pattern 31 and differentiatin features 32 in rder t classify the specimens. Five types f grwth pattern were recgnised (Fig. 1). Superficial in which multiple turnut peninsula extended frm the epidermis and/r adnexae and abutted r penetrated the papillary dermis. Ndular which cnsisted f a runded mass f tumur cells with a well defined cntur and peripheral pallisading. Fig. 1 Figure 1--Histlgical grwth patterns f basal cell carcinma (~cc). Micrndular in which the tumur islands tended t be smaller and mre numerus than the ndular type but still retained a unifrm cntur. Infiltrative, characterised by an irregular cntur t the tumur islands with a tendency t frm spikes. The turnut islands culd be large r small usually with prly develped peripheral pallisading. Mrpheic. In this subtype, the turnut islands were small and elngated with jagged ends and the strma is sclertic. The neplastic tissue infiltrates as narrw strands r crds f cells. We did nt define a mixed categry and chse t define the tumur by the predminant ppulatin. Superimpsed n t this classificatin was an assessment f differentiatin. The presence f cystic, adenid differentiatin r features f squamus metaplasia were cnsidered as evidence f differentiatin, therwise turnuts were classified as undifferentiated (slid). Ki-67 immunhistchemistry Sectins f 4gin were munted n ply-l-lysine cated slides and air dried at 37~ vernight. The sectins were deparaffinised, rehydrated and endgenus perxidase blcked using standard prtcls. Antigen retrieval was enhanced by fur micrwave irradiatin cycles f 5rain at 850mW in 10raM citric acid (ph 6.0). After standing fr 10 rain the sectins were washed, flded with nrmal swine serum fr 10 rain and incubated vernight at 37~ in a 1:200 dilutin f rabbit anti-ki-67 plyclnal antibdy (MIB-1, Dak Ltd, High Wycmbe, Bucks) in 0.01 M tris buffered saline (TBS) (ph7.4) After washing three times, a 1:300 dilutin f bitinylated swine anti-rabbit IgG (Dak Ltd) was applied fr 1 h, fllwed by a 1:80 dilutin f avidin-bitin perxidase cnjugate (Dak Ltd) fr a further hur. The sectins were visualised with Diaminbenzidine, cunterstained and munted. Assessment f Ki-67 labelling The percentage f cells staining psitively fr Ki-67 was measured using a grid n a TV mnitr cnnected

3 A cmparisn f the grwth fractin f hrrifying and nn hrrifying tumurs via a camera t the micrscpe. Ten high pwer (x 400) fields were cunted using alternate fields after r a n d m selectin f the starting psitin. T u m u r cells were identified n mrphlgical criteria and ttal and labelled cells cunted; typically per field. In the n n hrrifying grup ne sectin per t u m u r was cunted with a mean f 2300 cells per patient, in the hrrifying grup there were mre sectins available and a range f ne t three sectins were cunted with a mean f 3250 cells per turnut. The grwth fractin was determine as the n u m b e r f Ki67 psitive cells expressed as a percentage f the ttal n u m b e r f cells. The spatial distributin f prliferatin was assessed with reference t the arrangement f Ki-67 labelling within the t u m u r islands. Fur distinct patterns were classified (Fig. 2). 61 Marginal represented specimens in which prliferating cells were restricted t the peripheral cell layers f the t u m u r islands. Randm distributin was fund in m a n y t u m u r s which cnsisted f diffuse staining t h r u g h u t the t u m u r cells withut any bvius cncentratin f prliferating cells in the pallisading r ther cell layers. Marginal difjuse pattern where a tendency fr peripheral staining was c m b i n e d with a diffuse pattern t h r u g h u t the remainder f the t u m u r area which diminishes centrally. Mixed in which cmbinatins f tw r m r e f the distinct patterns c-existed within the t u m u r specimen. Fig. 2 Figure ~Phtmicrgraphs f prliferatin patterns in basal cell carcinma. (A) Marginal pattern in a ndular BCC with cystic differentiatin, ( 100). (B) Marginal diffuse pattern in a ndular BCC with cystic differentiatin ( 100). (C) Randm pattern in an infiltrative BCC (x 100). (D) Mixed pattern in a ndular BCC (x 100).

4 62 British Jurnal f Plastic Surgery Statistics The distributin f the Ki67 grwth fractin was psitively skewed (P = 0.04, W = 0.94). Square rt transfrmatin was therefre perfrmed t prduce a nrmal distributin (P=0.88, W=0.98)? 3 Cmparisns between the transfrmed grwth fractins were perfrmed using Student's t-test. Chi-squared gdness f fit analysis was utilised fr nminal data such as prliferatin pattern. Nminal 5% levels fr statistical difference was used thrughut. 95% cnfidence intervals fr different grups were calculated. Assuming a hypthesis f detecting 10% differences in grwth fractin between the grups, with a standard deviatin f 13, the study wuld have a pwer f 90% in detecting a difference accrding t the nmgram f detecting pwer frm sample size. 34 Results Patients A ttal f 22 patients with hrrifying tumurs and 81 nn hrrifying tumurs frm 78 patients were prcessed. The age, sex and relevant aetilgical factrs are in Table 1. THmu~ The site, size and number f recurrences are seen in Table 2. All f the tumurs were destructive and tw patients were cnsidered t have metastatic BCC. One patient develped lymph nde metastasis in the neck which was histlgically prven; the ther patient, wh had Grlin's syndrme develped an intracranial tumur fllwing an rbital exenteratin which at that time was cnsidered t be metastatic BCC. Presentatin Of the 22 patients with hrrifying turnuts, 11 presented late having neglected their turnut and nt sught medical advice despite being aware f its presence. This was mst cmmnly due t a fear f medical interventin. The patients admitted t having their tumurs fr 1-20 years (mean 8.2 years). The ther 11 patients presented early and develped hrrifying turnuts as a cnsequence f failure f treatment and the develpment f recurrences. Outcme Of the 11 patients wh presented late, 9 patients underwent extensive surgical resectin fllwed by Table 1 Patient factrs in the hrrifying and nn hrrifying grups Hrrifying Nn hrrifving Age years (Mean 62 years) years (Mean 70 years) Sex 12 M, 10 F 38 M, 40 F Aetilgical factrs 1 Grlin's syndrme 1 Grlin's syndrme 2 X-irradiatin in childhd Site 21 Head and neck 62 Head and neck 6 Scalp 23 upper 6 Orbit 22 middle 3 Periauricular/partid/mastid 12 lwer 6 Other 5 neck 1 Abdmen 10 Trunk 9 Legs Size Recurrences Destructive Metastasis Mean 10.1 cm, Range 5-25 cm 7 with mre than 4 recurrences Mean 2.7, range Patients 2 Patients Mean 0.9 cm, Range cm 0 Table 2 The distributin f histlgical subtypes in the hrrifying and nn hrrifying grups Histlgical Nn All Early presenting Late presenting grwth pattern hrrifying hrrifying hrr(/~ving hrrij):ing (81) (22) (11) (11) Micrndular Ndular Infiltrative Mrpheic Superficial 15

5 A cmparisn f the grwth fractin f hrrifying and nn hrrifying tumurs 63 recnstructin. Of these, eight patients had cmplete excisin and are cnsidered disease free at present (fllw-up range 1 10 years, mean 4.1 years). One patient had incmplete excisin fllwed by raditherapy. This patient has develped further recurrence and has cerebral invlvement frm direct extensin despite further surgery and raditherapy. Tw patients refused primary surgery and underwent raditherapy. Bth develped recurrences which were eventually managed surgically. One patient had cmplete excisin and is cnsidered disease free after extensive resectin. The ther had incmplete excisin resulting in further recurrence which has nt been cntrlled. Of the 11 patients presenting early, 7 had raditherapy primary treatment, fllwed by a recurrence. In all cases surgical excisin f this recurrence resulted in rerecurrence despite cmplete excisin (4 cases) r incmplete excisin (3 cases). Fllwing successive surgical resectins five patients have achieved lcal cntrl and tw patients cntinue t survive with prgressive disease. Fur patients had surgery at first presentatin. The patient with Grlin's had cmplete excisin but develped metastasis. The ther three had incmplete excisin (by curettage in ne patient) fllwed by raditherapy. Despite surgical treatment fr subsequent recurrence, tw never achieved lcal cntrl and the ther develped metastasis. Histlgical grwth pattern and differentiatin Nn hrrifying tumurs. The 81 tumurs were classified as 38 ndular, 7 micrndular, 15 infiltrative, 6 mrpheic and 15 superficial (Table 2). Within each f these sub-types there was diversity in the expressin f differentiatin features. Bth micrndular and mrpheic types failed t exhibit any differentiatin and their structure was a slid and cntiguus mass f cells. Hwever, ndular BCC shwed evidence f a variety f differentiated features: 24% were adenid, 18"/0 shwed cyst frmatin, 5% were kerattic and 5% shwed squamus differentiatin, leaving 47% as slid tumur masses with n distinguishing features. Bth superficial and infiltrating tumurs were predminantly undifferentiated althugh 15 20% shwed adenid features. Hrrifying turnurs. The specimens frm the first treatment f the early presenting tumurs were available in 7 f the 11 patients. The 2nd bipsy was available in three patients and the 3rd in ne patient. All f the late presenting grup had specimens available fr examinatin. The hrrifying tumurs were classified as 12 infiltrative, 4 micrndular, 1 mrpheic and 5 ndular (Table 2). N differences were evident in the histlgical grwth pattern f early and late presenting tumurs (Table 2). The turnuts in which multiple bipsies were available tended t maintain their riginal grwth pattern apart frm ne micrndular and ne ndular tumur which develped an infiltrative grwth pattern n recurrence. Six tumurs exhibited squamus metaplasia, three were cystic, five shwed adenid features and nine were cnsidered undifferentiated. Grwth fractin. The mean and median grwth fractin fr the primary BCC were 29.3% and 28.1% respectively and ranged frm 6.4 t 76.9%. The first available bipsy all f the hrrifying tumurs had a mean and median f 30.9% and 31.8% respectively with a range f 8.4% t 62.3%. Figure 3 shws that the distributin f the Ki-67 grwth fractin was similar between the 2 grups. Statistical analysis f the grwth fractins f hrrifying and nn hrrifying grups did nt demnstrate a significant difference. The estimated difference f the square rt transfrmed grwth fractin was with a 95% cnfidence interval f t (Student's t-test P = 0.8). The grwth fractin f sequential specimens frm thse hrrifying tumurs with multiple recurrences were als cmpared t establish whether prliferatin altered with tumur prgressin (Fig. 4). The differences were nt transfrmed because they were nrmally distributed (P = 0.07, W = 0.85). 33 There was n 0.8- O 0,7_ g Ia_ O ( O D [] [] :~ ~ a Nn hrrifying Hrrifying Early presenting Early presenting Late presenting First Bipsy Last Bipsy Patient Grup Fig. 3 Fig. 4 Figure 3--A cmparisn f the grwth fractin f hrrifying and nn hrrifying tumurs. Figure 4~A graph demnstrating that (A) the grwth fractin des nt alter with tumur prgressin. (B) There is n difference in the grwth fractin f hrrific tumurs that present late r that develp in spite f early interventin.

6 64 British Jurnal f Plastic Surgery Table 3 The grwth fractins f hrrifying and nn hrrifying basal cell carcinma f similar histlgical subtype are nt significantly different. Infiltrative tumurs in either grup have a higher grwth fractin than ndular tumurs Histlgical Hrrifying Nn hrrijying Estimated difference 95% P value. grwth pattern (mean grwth fractin) (mean grwth fractin) (transfrme data) cnfidence interval (Student's t-test f (transfrmed data) transfrmedata) Infiltrative 37.4% 37.3% t Ndular 15.3% 24.9% ~ t Micrndular 30.1% 23.2% t difference between the grwth fractin f the first r last bipsies f tumurs with serial specimens. The estimated difference f the grwth fractin was with a 95% cnfidence interval f t (Paired t-test P = 0.5). The cnfidence interval is wide due t the small number f specimens available with sequential bipsies. Figure 4 als shws the cmparisn f Ki-67 grwth fractin between thse tumurs which presented early and became hrrifying and thse that presented late due t neglect. The first available bipsies f the early presenting tumurs were nt different t the late presenting tumurs. The estimated difference f the transfrmed grwth fractin was with a 95% cnfidence interval f t (Student's t-test P = 0.77). In rder t determine whether there may be a difference in the bilgy f tumurs f similar histlgical grwth pattern in the hrrifying and nn hrrifying grups the grwth fractin in each subtype was cmpared (Table 3). There was n difference between the infiltrative, ndular r micrndular tumurs in each grup. In agreement with ther findings 35 the grwth fractin f the infiltrative tumurs was significantly higher than the ndular tumurs fr bth the series f nn hrrifying patients and hrrifying patients. (Nn hrrifying; estimated difference 1.21, 95% cnfidence interval 0.06 t 0.19, P = Hrrifying; estimated difference 0.21, 95% cnfidence interval 0.08 t 0.34, P = ) The prliferatin pattern. In primary basal cell carcinma the spatial distributin f the prliferating ceils was described in fur distinct patterns. In the nn hrrifying tumurs the mst cmmn discrete patterns were diffuse, which accunted fr 42% f tumurs, and the marginal diffuse pattern which was fund in 31% f the specimens. The marginal staining pattern was nly evident in tw tumurs whilst 25% had a mixture f staining patterns, in the hrrifying grup 63% exhibited a diffuse pattern, 18% a marginal diffuse pattern, 14~ a mixed pattern and nly ne tumur stained with a marginal pattern. There was n significant difference between the prliferatin patterns f the 2 grups. (Chi squared gdness f fit test 3.93, 3 degrees f freedm, P = 0.27.) Discussin When Arthur Jacb first described BCC in 1827 as, 'a destructive ulcer f peculiar character which I bserved t attack the eyelids, and extend t the eye-ball, rbit, and face' he was describing what may be defined as a hrrifying tumur. Presently hrrifying tumurs are rare and there are few published series examining bilgical features which may accunt fr the develpment f this type f tumur. Sme publicatins relate t management techniques, thers describe clinical and histpathlgical cnsideratins althugh usually in small numbers f patients. 24 6~.,,, 12,40 42 Hwever, few examine the bilgical prcesses that might determine an aggressive r hrrifying tumur phentype. The questin remains as t whether all BCCs culd ptentially becme hrrifying if left unchecked r whether sme tumurs pssess certain bilgical features that determine the develpment f this particularly aggressive tumur phentype. Sme papers suggest they are histlgically indistinguishable, ~~ thers implicate reduced hst immunity "',~4 r expsure t x-irradiatin as imprtant aetilgical factrs? 6 The predminant histlgical difference assciated with aggressive tumurs has been the prevalence f an infiltrative grwth pattern. 2,43 Recent studies have demnstrated altered p53 and bcl-2 ncprtein expressin, strmal and nuclear differences and altered nevascularizatin f infiltrative r histlgically aggressive tumurs cmpared t nn aggressive tumurs These studies, hwever, relate t mrphlgically aggressive tumurs rather than t a clinical grup f patients with hrrifying r aggressive BCC. It remains equivcal whether these differences imply that hrrifying tumurs result frm a bilgically mre aggressive subtype r the fact that certain types, especially infiltrative tumurs, are mre cmmnly incmpletely excised and the changes result frm tumur prgressin. In this series the majrity f tumurs (and all f the tumurs which became hrrific by way f incmplete excisin and recurrence) were situated n the head and neck, an area in which excisin margins may be cmprmised fr csmetic purpses. Of fur patients wh presented early and were treated by primary surgery, three were reprted t have incmplete excisin. The furth patient was unusual in having Grlin's syndrme and develped a metastasis fllwing cmplete excisin. Althugh the numbers are small, they supprt the cncept that early inadequate treatment may be an imprtant aetilgical factr. Hwever, it has been reprted that nly 42~ f the inadequately excised tumurs will recur in the shrtterm suggesting that bilgical characteristics play an imprtant rle. 47 In agreement with ther publicatins 13 (60%) exhibited an infiltrative r mrpheic grwth pattern

7 A cmparisn f the grwth fractin f hrrifying and nn hrrifying tumurs 65 initially, this increased t 15 (70%) with subsequent recurrence and tumur prgressin. Hwever, the grwth pattern alne cannt be implicated as the sle cause f develpment f hrrifying tumurs since in a nrmal cnsecutive series 7% wuld exhibit an infiltrative grwth pattern 3' and nt all these becme hrrific. Five f the hrrifying tumurs (23%) were ndular which is traditinally cnsidered t be the less aggressive frm f BCC. Three f these ndular tumurs presented at an early stage and became recurrent despite treatment. This suggests that all histlgical subtypes may ptentially becme hrrifying and that there may be additinal factrs cupled with histlgy that play an imprtant rle in determining the behaviur f these turnuts. Cellular prliferatin is a fundamental determinant f tumur grwth behaviur and unchecked cell divisin facilitates the acquisitin f further genetic mdificatin which may include the prpensity t infiltrate and metastasise. '7 The grwth fractin in BCC has been shwn t exhibit wide variatin and recurrent tumurs have been shwn t express a higher grwth fractin independent f histlgical subtype. It was cnceivable that the grwth fractin may determine the behaviur f these tumurs indicating the existence f an aggressive subset f BCC. Hwever, the grwth fractin, measured by Ki 67 immunhistchemistry, did nt supprt this hypthesis. The grwth fractin r pattern f cellular prliferatin f 22 hrrifying tumurs was nt significantly different t a grup f 81 nn hrrifying turnuts f a mixed histlgical subtype. Furthermre, there was n difference in grwth fractin f hrrifying and nn hrrifying tumurs f similar histlgical grwth pattern (Table 3). This suggests that infiltrative r ndular tumurs f hrrifying and nn hrrifying tumurs are bilgically similar. There was, hwever, a trend that hrrifying ndular lesins were less prliferative than nn hrrifying lesins. A cmbinatin f high grwth fractin and histlgical grwth pattern culd nt predict prblem lesins. These findings suggest that ndular, micrndular r infiltrative BCCs that becme hrrifying are nt different t thse that fllw a rdinary curse. The finding that infiltrative tumurs in either hrrifying r nn hrrifying grups have a higher grwth fractin is highly significant and supprts the view that these tumurs are bilgically different and may cntribute t their mre invasive behaviur resulting in a higher incmplete excisin rate. It suggests that failure f primary treatment determines whether these tumurs prgress t becme hrrifying rather than there being an intrinsic bilgical difference between nn hrrifying and hrrifying tumurs f similar histlgical grwth pattern. Late presentatin and tumur neglect has been implicated as cmmn feature f mutilating lesins. In this series nly half f the patients presented late. The ther half f patients presented early with nn hrrific tumurs which recurred and became hrrific as the tumur prgressed. It wuld nt be unreasnable t expect that the bilgy f the tumurs f these 2 grups may be different. The early presenting grup cnsisting f inherently aggressive tumurs which were resistant t treatment, and the late presenting grup cnsisting f less aggressive tumurs. Histlgically, there was a similar mix f histlgical grwth patterns in each grup with the infiltrating type equally divided (Table 2) and there was n difference in the grwth fractin between the early and late presenting grups. Clearly this des nt supprt these expectatins f differing bilgies. The clinical histries f the early presenting grup reveal that three f the fur patients managed surgically had incmplete excisin initially, which may explain their recurrence and further recurrences. The ther seven patients develped recurrences fllwing primary raditherapy. Further recurrence resulted fllwing surgical excisin despite cmplete r incmplete excisin margins. It is unclear whether this clinical data reflects either a subgrup f tumurs with particularly aggressive tumur phentype r that previus failed therapy disturbed tissue planes hindering subsequent surgical excisin and clear definitin f histlgical margins. The cnsequence f the latter may allw prgressive tumur grwth int deeper and less accessible tissue planes. An alternative explanatin wuld be that gradual tumur prgressin, with successive recurrence, may result in a mre aggressive phentype. Hwever analysis f sequential specimens frm 12 patients with recurring tumurs failed t demnstrate a change in grwth fractin. This des nt supprt a cncept that recurrence and tumur prgressin results in a bilgically mre aggressive tumur. In cnclusin, this study des nt supprt the cncept that hrrifying tumurs are bilgically different t nn hrrifying tumurs. It suggests that many BCCs (nt exclusively f infiltrative subtype) have the ptential t behave in this aggressive fashin. In half f the patients neglect was an imprtant factr. Hwever there remain a prprtin f patients wh present early and still develp hrrifying tumurs. These appear t be n different t the late presenting tumurs r t nn hrrifying tumurs f similar histpathlgical subtype. Whether the difference in clinical behaviur f these tumurs frm nn hrrifying tumurs represents unidentified bilgical factrs, r mre likely, the failure f early f management, especially f infiltrating tumurs, remains unclear. Early effective primary treatment wuld appear t be essential t prevent the prgressin f basal cell carcinma t a hrrific clinical type. Acknwledgements The wrk was supprted by the Restratin f Appearance and Functin Trust. The authrs wuld like t thank RAFT and the Gray labratry fr their help. References 1. Histlgical typing f skin tumurs. Geneva: Wrld Cnventin; Wrld Health Organizatin. 2. Jacbs GH, Rippey J J, Altini M. Predictin f aggressive behavir in basal cell carcinma. Cancer 1982; 49: Rwe DE, Carrll RL Day CL, Jr. Lng-term recurrence rates in previusly untreated (primary) basal cell carcinma: implicatins fr patient fllw-up. J Dermatl Surg Oncl 1989; 15:

8 66 British Jurnal f Plastic Surgery 4. Vic R Furez T, Nemec E, Andry G, Deraemaecker R. Aggressive basal cell carcinma f head and neck areas. Eur J Surg Oncl 1995; 21: McGurk M, Edwards MB. Mutilating basal cell carcinma f the face. Br J Oral Maxillfac Surg 1984; 22: Dvretzky l, Fisher BK, Haker O. Mutilating basal cell epithelima. Arch Dermatl 1978; 114: Schwartz RA, Vickerman CE, Phelan JT, et al. Mutilating sclersing basal cell epithelima. J Surg Oncl 1979; 12: Bianchini R, Wlter M. Fatal utcme in a metatypical giant, "hrrifying" basal cell carcinma. J Dermatl Surg Oncl 1987; 13: Schwartz RA, De Jager RL, Janniger CK, Lambert WC. Giant basal cell carcinma with metastases and myelphthisic anemia. J Surg Oncl 1986; 33: Jacksn R, Adams RH. Hrrifying basal cell carcinma: a study f 33 cases and a cmparisn with 435 nn-hrrr cases and a reprt n fur metastatic cases. J Surg Oncl 1973; 5: Leffell D J, Headingtn JT, Wng DS, Swansn NA. Aggressivegrwth basal cell carcinma in yung adults. Arch Dermatl 1991; 127: Itayemi SO, Abiye AA, Ogan O, Daramla JO, Oluwasanmi JO. Aggressive basal cell carcinma in Nigerians. Br J Dermatl 1979; 101: Randle HW, Renigk RK, Brdland DG. Giant basal cell carcinma (T3). Wh is at risk? Cancer 1993; 72: Weimar VM, Ceilley RI, Geken JA. Aggressive bilgic behavir f basal- and squamus-cell cancers in patients with chrnic lymphcytic leukemia r chrnic lymphcytic lymphma. J Dermatl Surg Oncl 1979; 5: Oram Y, Oreng I, Grieg RD, Rsen T, Thrnby J. Histlgic patterns f basal cell carcinma based upn patient immunstatus. Dermatl Surg 1995; 21: Spr H J, Lind SD, Erlandsn RA. Skin cancer: cmparisn f aggressive ptentials f sunlight-induced and x-ray-induced tumrs. Cutis 1977; 20: Tubiana M, Curdi A. Cell prliferatin kinetics in human slid tumrs: relatin t prbability f metastatic disseminatin and lng-term survival. Radither Oncl 1989; 15: Cattretti G, Pileri S, Parravicini C, et al. Antigen unmasking n frmalin-fixed, paraffin-embedded tissue sectins [see cmments]. J PathI 1993; 171: Vernese SM, Gambacrta M, Gttardi O, Scanzi F, Ferrari M, Lampertic R Prliferatin index as a prgnstic marker in breast cancer. Cancer 1993; 71: Sahin AA, R J, R JY, et al. Ki-67 immunstaining in ndenegative stage I/II breast carcinma. Significant crrelatin with prgnsis. Cancer 1991; 68: Miller TP, Grgan TM, Dahlberg S, et al. Prgnstic significance f the Ki-67-assciated prliferative antigen in aggressive nn-hdgkin's lymphmas: a prspective Suthwest Onclgy Grup trial. Bld 1994; 83: Yamanaka N, Harabuchi Y, Kataura A. The prgnstic value f Ki-67 antigen in nn-hdgkin lymphma f Waldeyer ring and the nasal cavity. Cancer 1992; 70: Yussef EM, Matsuda T, Takada N, et al. Prgnstic significance f the MIB-1 prliferatin index fr patients with squamus cell carcinma f the esphagus. Cancer 1995; 76: Ramsay JA, Frm L, Isce NA, Kahn HJ. MIB-1 prliferative activity is a significant prgnstic factr in primary thick cutaneus melanmas. J Invest Dermatl 1995; 105: Baum HR Meurer I, Unteregger G. Ki-67 antigen expressin and grwth pattern f basal cell carcinmas. Arch Dermatl Res 1993; 285: Healy E, Angus B, Lawrence CM, Rees JL. Prgnstic value f Ki67 antigen expressin in basal cell carcinmas. Br J Dermatl 1995; 133: Weinstein GD, Frst R Cell prliferatin in human basal cell carcinma. Cancer Res 1970; 30: Yshimura K, Egawa K, On T. [Prliferating patterns f basal cell carcinma - immunhistlgical study by means f anti- BrdU mnclnal antibdy]. In Japanese. Nippn Hifuka Gakkai Zasshi (Japanese Jurnal f Dermatlgy) 1990; 100: Grimwd RE, Ferris CF, Mercill DB, Huff JC. Prliferating cells f human basal cell carcinma are lcated n the periphery f tumr ndules. J Invest Dermatl 1986; 86: al Sader MH, Dyle E, Kay EW, et al. Prliferatin indexes - a cmparisn between cutaneus basal and sqaamus cell carcinmas. J Clin Pathl 1996; 49: Sextn M, Jnes DB, Malney ME. Histlgic pattern analysis f basal cell carcinma. Study f a series f 1039 cnsecutive neplasms. J Am Acad Dermatl 1990; 23: Lever WE Pathlgy f cmmn skin tumrs. J Surg Oncl 1971; 3: Shapir SS, Wilk MB. An Analysis f Variance Test fr Nrmality (Cmplete samples). Bimetrika 1965; 52:519~ Chapman DG. Practical Statistics fr medical research: Chapman and Hall, 1991: Hrlck NM, Wilsn GD, Daley FM, Richman PI, Sanders R. Eur J Oncl 1997; 23: Stenquist B, Wennberg AM, Gisslen H, Lark O. Treatment f aggressive basal cell carcinma with intralesinal interfern: evaluatin f efficacy by Mhs surgery. J Am Acad Dermatl 1992; 27: Ple MD, Cchrane TD, Bwen JE. Aggressive nn-irradiated basal cell carcinma f frehead treated by free flap transfer. Br J Plast Surg 1975; 28: Mnheit GD. Micrscpically cntrlled surgery: Mbs' chemsurgery fr recurrent and aggressive skin cancer. Ala J Med Sci 1981; 18: Albright SD3d. Case reprt: prlnged cure f extensive primary and recurrent cancers f the skin by aggressive crysurgery. J Dermatl Surg Oncl 1983; 9: de Rsa G, Vetrani A, Zeppa R et al. Cmparative mrphmetric analysis f aggressive and rdinary basal cell carcinma f the skin. Cancer 1990; 65: Grmley DE, Hirsch R Aggressive basal cell carcinma f the scalp. Arch Dermatl 1978; 114: Jernbeck J, Glaumann B, Glas JE. [Basal cell carcinma. Clinical evaluatin f the histlgical grading f aggressive types f cancer]. (In Swedish) Lakartidningen 1988; 85: Slane JR The value f typing basal cell carcinmas in predicting recurrence after surgical excisin. Br J Dermatl 1977; 96: De Rsa G, Barra E, Guarin M, Staiban S, Dnfri V, Bscain A. Fibrnectin, laminin, type IV cllagen distributin, and myfibrblastic strmal reactin in aggressive and nnaggressive basal cell carcinma. Am J Dermatpathl 1994; 16: De Rsa G, Staiban S, Barra E, Dnfri V, Salvatre G, Vessecchia G, et ai. p53 prtein in aggressive and nnaggressive basal cell carcinma. J Cutan Pathl 1993; 20: De Rsa G, Staiban S, Barra E, et al. Nuclelar rganizer regins in aggressive and nn-aggressive basal cell carcinma f the skin. Cancer 1992; 69: Delln AL, DeSilva S, Cnnlly M, Rss A. Predictin f recurrence in incmpletely excised basal cell carcinma. Plast Recnstr Surg 1985; 75: 86~71. The Authrs Nigel M. Hrlck MB, BS, FRCS, Research Fellw, RAFT Institute f Plastic Surgery. Gerge D. Wilsn BSc, PhD, Senir Research Scientist, Gray Labratry Cancer Research Trust. Francis M. Daley MSc, Gray Labratry Cancer Research Trust Paul I. Richman MB, BS, PhD, FRC (Path), Department f Histpathlgy Adriaan O. Grbbelaar MMed (Plast), FCS (SA), FRCS (Plast), Cnsultant Plastic Surgen Ry Sanders BSc, FRCS, Cnsultant Plastic Surgen Christpher Fy MSc, Statistician, Department f Research and Develpment, Munt Vernn Hspital, Nrthwd, Middlesex, HA6 2RN, UK. Crrespndence t Mr Nigel M. Hrlck. Paper received 29 January Accepted 30 September 1997, after revisin.