Endocrine-Related Cancer (2002) 9 103–113

A critical reappraisal of MIB-1 labellingindex significance in a large series ofpituitary tumours: secreting versusnon-secreting adenomas

M L Jaffrain-Rea 1,2, D Di Stefano 3, G Minniti 4, V Esposito 4,A Bultrini 1, E Ferretti 1,2, A Santoro 5, L Faticanti Scucchi 3,A Gulino 3 and G Cantore 4,5

1Department of Experimental Medicine, Universita degli Studi di L’Aquila, Via Vetoio, Coppito 2, 67100 L’Aquila(AQ), Italy2Fondazione Carlo Ferri per la Prevenzione e la Cura dei Tumori, Via E Riva 42, 00015 Monterotondo (RM),Italy3Department of Experimental Medicine, Universita La Sapienza, Viale Regina Margherita, 00161 Rome (RM),Italy4Neuromed Institute, IRCCS, Pozzilli (FR), Italy5Department of Neurological Sciences, Universita La Sapienza, Viale Regina Margherita, 00161 Rome (RM),Italy

(Requests for offprints should be addressed to M-L Jaffrain-Rea, Dipartimento di Medicina Sperimentale,Universita degli Studi di L’Aquila, Via Vetoio, Coppito 2, 67100 L’Aquila (AQ), Italy; Email:jaffrain.ml.rea@katamail.com)

Abstract

Pituitary tumours are usually benign neoplasia, but may have a locally aggressive or malignantevolution. This study aimed to identify factors which mostly influence their proliferative activity, inorder to clarify its value for clinical and research purposes. The proliferative index was determined ina prospective series of 132 pituitary tumours as the percentage of monoclonal antibodyMIB-1-immunopositive cells and referred to as the MIB-1 labelling index (LI). Its distribution wasanalysed according to both univariate and multivariate models. A life-threatening pituitary tumour ispresented separately. The mean LI was 1.24 ± 1.59%, with significant differences between clinicallysecreting (CS) and clinically non-secreting (CNS) adenomas. In CS adenomas (n = 65), LI was highlyvariable and markedly influenced by pre-operative pharmacological treatment (0.80 ± 1.03 vs2.06 ± 2.39% in treated vs untreated cases, P = 0.009); it decreased with patient’s age (P = 0.025, r =0.28) and increased with tumour volume and invasiveness. The influence of pre-operative treatmentand macroscopic features on LI in this group was confirmed by multivariate analysis. In CNSadenomas (n = 67), LI distribution was less variable than in CS adenomas (P < 0.0001), it wasage-independent and correlations with tumour volume, invasiveness or recurrence did not reachsignificance. In a rapidly growing parasellar tumour, the mean LI was 24% at first surgery andexceeded 50% at second surgery performed 4 months later. LI should be interpreted according tohormone secretion and pre-operative treatment. Unusually high LI values deserve particular attention.

Endocrine-Related Cancer (2002) 9 103–113

IntroductionAlthough pituitary tumours are usually benign and slowlygrowing neoplasms, their evolution is highly variable, with abroad clinical spectrum ranging from apparently steadymicroroadenomas to grossly invasive or exceptionally malig-nant metastatic tumours. Their management is based on a

Endocrine-Related Cancer (2002) 9 103–113 Online version via http://www.endocrinology.org1351-0088/02/009–103 2002 Society for Endocrinology Printed in Great Britain

multidisciplinary approach which includes surgery, radio-therapy and pharmacological therapy with either somatosta-tin analogues or dopamine agonist (DA) drugs for growthhormone (GH)- and/or prolactin (PRL)-secreting adenomas.Whereas DAs exert a direct anti-mitotic effect on PRL-secreting cells (Bevan et al. 1992), the cytostatic effect of

M L Jaffrain-Rea et al.: MIB-1 in pituitary tumours

somatostatin (SMS) analogues is still controversial(Lamberts et al. 1996, Thapar et al. 1997b). With the excep-tion ofprolactinomas, surgery remains a first-line therapeuticchoice for most pituitary adenomas, the indications for post-operative radiotherapy being still under debate (Jaffrain-Reaet al. 1993, Plowman 1995). Since few correlations havebeen observed between clinical evolution and histopatholog-ical features, highly aggressive or malignant tumours mayinitially present as common macroadenomas, and reliablemarkers of cell proliferation are expected to be a useful toolfor their therapeutic management.

The pathogenesis of pituitary tumours is currently underextensive investigation, and molecular events closely interactwith endogenous/exogenous permissive factors to determinethe expansion of monoclonal cell populations (Melmed1999). Markers of cell proliferation are also attractive forresearch purposes, by helping to identify factors associatedwith regulation of tumour growth and/or kinetic activity(Thapar et al. 1997b, Shimon et al. 1998, Tampanaru-Sarmesiu et al. 1998, Yokoyama et al. 1998, Jaffrain-Rea etal. 1999, Nishi et al. 1999, Turner et al. 2000). However,possible pitfalls in its interpretation should be carefully con-sidered.

The MIB-1 antibody recognises the Ki-67 antigen, whichis expressed throughout the cell cycle, and the percentage ofimmunopositive cells represents a reliable proliferation indexfor the study of human tumours, usually referred to as theMIB-1 labelling index (LI). Because of its applicability toroutinely fixed paraffin-embedded sections, the MIB-1 anti-body has now replaced the previous anti-Ki-67 antibodies(Cattoretti et al. 1992). The first reports on the use of Ki-67and MIB-1 antibodies in pituitary tumours were published in1986 (Landolt et al. 1987) and 1995 (Ekramullah et al. 1995,Gandour-Edwards et al. 1995, Kawamoto et al. 1995)respectively. Yet, despite a wide literature accumulating onthis topic, the interpretation of LI in pituitary tumoursremains surprisingly conflicting under many aspects (forreviews see Losa et al. 1998, Turner & Wass 1999). Thismight be explained by various factors, among which the mostimportant may be the following: (i) most papers have focusedon a few specific characteristics of the studied tumours, someof them quite heterogeneous, often leading to apparently dif-ferent conclusions concerning LI variations according totumour volume/invasivity, recurrence or secreting activity,(ii) the possible influence of age or pre-operative pharmaco-logical treatment was generally not considered, and (iii) onlyLI mean values were considered, regardless of LI variability,which represents another important feature of LI distri-bution.

Thus, the aim of the present study was to better definethe distribution and biological significance of LI in pituitarytumours by taking into account all the parameters cited hith-erto in a large prospective series, using univariate models tocompare data with previous studies and multivariate models

104 www.endocrinology.org

to critically analyse these results and explain current discrep-ancies.

Materials and methods

Patients

A hundred and thirty-two consecutive pituitary tumours wereoperated on over a 4 year period and prospectively studied.Surgery was performed by either a transphenoidal or a trans-cranial route, 12 tumours being re-operated on during thestudy period because of tumour regrowth. Detailed clinical,hormonal and radiological information were obtained in allcases before surgery. Clinically secreting (CS) tumours werediagnosed according to currently accepted basal and dynamichormonal criteria, the lack of bioclinical feature of hormonehypersecretion defining clinically non-secreting (CNS)tumours. Sixty-eight tumours were classified as CNS,whereas 42 presented with acromegaly, six with Cushing’sdisease, 15 were PRL-secreting and one was thyrotrophin(TSH)-secreting. At the time of surgery, 31 patients were onmedical treatment for CS tumours: 15 were receiving DAdrugs for PRL-secreting (n = 9) and/or GH-secreting (n = 6)tumours and 20 were receiving s.c. octreotide (OCT) foractive acromegaly (in association with DA drugs in fourcases). Only two patients with recurrent CNS tumours hadreceived radiotherapy, but seven patients with CNS hadreceived a pre-operative course of DA drugs because of asso-ciated hyperprolactinaemia. Pre-operative NMR imaging wasavailable in all cases. Microadenomas, defined by maximaltumour diameter �10 mm, were present in only 14 cases(eight with acromegaly, five with Cushing’s disease and oneCNS). Macroscopic features of invasiveness were definedaccording to pre-operative radiological criteria and intra-operative findings, including macroscopic evidence for duralinfiltration. According to such criteria, 48 adenomas were non-invasive and 84 were invasive. Tumours were also classifiedaccording to their suprasellar extension (SSE) using Wilson’scriteria (Wilson 1984): no SSE (grade O; n = 28), moderateSSE (grade A/B; n = 68), huge SSE (grade C/D; n = 36).Invasion of the cavernous sinus (grade E according to Wilson)and of the sphenoid sinus were recorded separately.

Archive material from five patients (four with CNS andone with acromegaly) who were operated on for tumourrecurrence during the study period but had previously under-gone surgery at the same Neurosurgery Unit was also col-lected for retrospective determination of the LI. These caseswere used only for the study of recurrences.

During the prospective period of study, a 33-year-oldmale patient was also operated on for a parasellar mass whichwas finally diagnosed as a probable follicle-stimulating hor-mone (FSH)-secreting pituitary carcinoma. Data arepresented as a separate case report and excluded from thestatistical analysis.

Endocrine-Related Cancer (2002) 9 103–113

Tumours

Immunohistochemical determination of pituitary tumour hor-monal content was performed on paraffin-embedded tissuesections with polyclonal rabbit antibodies (anti-PRL,anti-GH, anti-FSH, anti-LH, anti-TSH, anti-ACTH) obtainedfrom Dako LSABZ System, HRP, Dako Corporation, Car-pinteria, CA, USA), using the streptavidin–biotin detectionsystem. According to such analysis, CNS tumours weremainly represented by null cell adenomas (n = 41) and gona-dotroph adenomas (n = 17), the others showing silent ACTHsecretion (n = 5) or scattered GH-secreting cells (n = 5). Outof the 42 tumours associated with acromegaly, 21 were pureGH-secreting, 12 were mixed GH/PRL-secreting and sevenwere mixed GH/glycoprotein-secreting adenomas. Multihor-monal adenomas were observed in two tumours associatedwith acromegaly, the one associated with TSH-dependenthyperthyroidism and one prolactinoma. The remaining pro-lactinomas and all those associated with Cushing’s diseasedisplayed pure PRL- and ACTH-immunopositivity, respect-ively.

Evaluation of cell proliferation

The mouse monoclonal antibody (MoAb) MIB-1 (DiagnosticBrokers Associated, Milan, Italy) was used after microwavepretreatment, according to the manufacturer’s instructions.For each sample, immunopositivity was counted on 1000randomly selected cells by a single observer who wasunaware of the bioclinical characteristics of the tumours, andthe results were expressed as a percentage of positive cells.

In a representative samples of 40 primary tumours (15CNS, 25 GH- and/ or PRL-secreting adenomas, including 13treated and 12 untreated), morphological evaluation ofnuclear atypia and mitotic index (MI) was also carefully per-formed on haematoxylin–eosin stained sections. Nuclearatypia were classified as absent, mild, moderate or markedand the MI was determined on 1000 randomly selected cellsand expressed as a percentage.

Statistical analysis

Unless otherwise specified, all values are indicated asmeans ± s.d. The Mann–Whitney U-test and Kruskall–Wallistest were used as appropriate for univariate analysis ofunpaired data, and the Wilcoxon ranked test was used forpaired comparisons in the study of regrowing tumours. Vari-ations in LI distribution among subgroups were studied bythe F-test. Multivariate analysis were performed by MAN-COVA. The chi-square test was used to compared percent-ages. Correlations between the LI and the age of the patientswere studied by linear and multiple regression. P < 0.05 wasconsidered significant.

www.endocrinology.org 105

Results

Preliminary results

In order to identify possible bias in the interpretation of LIin this series, a careful preliminary study was performed.Special attention was paid to the comparison of CS and CNSadenomas (Table 1), which showed that patients with CSadenomas were younger (P < 0.0001) and presented less vol-uminous and invasive tumours than those with CNS aden-omas (P < 0.001 in both cases). In multivariate analysis,tumour volume and invasiveness were consistently found tobe strongly linked parameters, but independent from patient’sage and/or pre-operative treatment on the whole series (datanot shown).

General results

The mean LI in the whole series was 1.24 ± 1.59%, with amarked variability (see Fig. 1 and Table 2). Pre-operativetreatment with either DA or SMS drugs was found to sig-nificantly influence mean LI values (0.73 ± 0.97 vs1.45 ± 1.74% in treated vs untreated tumours respectively,P = 0.013) and to markedly reduce its variability(P < 0.0001). Accordingly, LI distribution was significantlyinfluenced by the functional characteristics of the tumoursand differed between CNS, treated and untreated CS (P =0.016) (Fig. 2). An inverse relationship was also foundbetween LI and the age of the patients in the whole series(P = 0.021, r = 0.20, Fig. 3), a similar trend being observedin untreated cases only (P = 0.050, r = 0.20).

Factors influencing LI distribution have thus been firstevaluated by univariate analysis in untreated pituitary aden-omas only (n = 94) (Table 3). Mean LI values tended toincrease with tumour volume, but only its variability wassignificantly higher in macroadenomas and especially in hugetumours. Mean LI values were higher in invasive than innon-invasive adenomas (P = 0.04), a similar difference beingfound between tumours invading the cavernous sinus com-pared with those which did not (P = 0.035), whereas invasionof the sphenoid sinus had no influence on this parameter.No significant difference was found between recurrent andprimary tumours. Mean LI values tended to be higher in CScompared with CNS adenomas (P = 0.092) and their varia-bility was markedly greater in the former group (P < 0.0001).Significant differences were found among the different hor-mone-secreting groups (P = 0.023), the highest values beingobserved in PRL-secreting adenomas (Fig. 4).

In a subsequent multivariate analysis taking into accountboth age and pre-operative pharmacological treatment, LIwas confirmed to be higher in macro- than in microadenomas(P = 0.013), in invasive than in non-invasive adenomas(P = 0.023), especially in those invading the cavernous sinus(P = 0.006), and to increase with the grade of SSE (P =0.001). It was also confirmed to be correlated with patient’s

M L Jaffrain-Rea et al.: MIB-1 in pituitary tumours

Table 1 Clinical characteristics of clinically secreting (CS) and non-secreting (CNS) pituitary adenomas

Cavernous SphenoidalAge Macro- sinus sinus

M/F (years) adenomas Invasive invasion invasionPituitary adenomas n ratio (means ± S.D.) (%) (%) (%) (%)

CNS adenomas 67 1.50 54.6 ± 12.2 98.6 76.1 58.2 22.3CS adenomas 65 0.91 43.5 ± 15.2 78.4 50.8 36.9 10.7

P (chi-square value) 0.22 1.46 <0.0001 0.0003 (13.2) 0.0025 (9.16) 0.027 (4.90) 0.067 (3.36)

Significant P values are underlined.

Figure 1 Labelling index (LI) in 132 consecutive pituitaryadenomas.

Table 2 MIB-1 labelling index (LI) distribution in a series of132 human pituitary adenomas. LI is expressed as apercentage of immunopositive cells

Distribution inpercentilesPituitary

adenomas n Means ± S.D. 10° 50° 90°

Whole series 132 1.24 ± 1.59 0.00 0.65 3.00Untreated 94 1.45 ± 1.74 0.00 1.00 3.18Treated 38 0.73 ± 0.97°,* 0.00 0.35 2.52

Secretingadenomas 65 1.44 ± 1.95 0.00 0.50 3.60Untreated 33 2.06 ± 2.39 0.08 1.00 6.00Treated 32 0.80 ± 1.03°,** 0.00 0.45 2.70

CNS adenomas 67 1.05 ± 1.12 0.00 0.80 2.46

°P < 0.0001 by F-test vs untreated patients, *P = 0.013 and**P = 0.09 by Mann–Whitney vs untreated patients.

106 www.endocrinology.org

Figure 2 LI in pituitary adenomas according to their func-tional characteristics. CNS: clinically non-secreting aden-omas; CS(t): clinically secreting adenomas, pre-operativetreatment with pharamacological drugs; CS(ut): clinicallysecreting adenomas, untreated. Means ± S.D. P<0.02 byKruskall-Wallis. *Significant difference between CNS, CS(t)and CS(ut).

Figure 3 Age-related distribution of the LI in 132 consecutivepituitary adenomas.

Endocrine-Related Cancer (2002) 9 103–113

Table 3 Factors influencing LI values in a series of 94 untreated pituitary adenomas. The number of cases is indicated withinbrackets for each subgroup. SSE = grade of suprasellar extension according to Wilson (1984)

LI (%) Mann–Whitney(means ± S.D.) ( + )

Gender, M (53) vs F (41) 1.58 ± 1.86 vs 1.29 ± 1.58 0.354

VolumeMicro- (8) vs macroadenomas (86) 0.60 ± 0.70 vs 1.53 ± 1.79°° 0.108SSE grade: O (14) vs A/B (48) vs C/D (30) 0.89 ± 0.90 vs 1.14 ± 1.36 vs 2.12 ± 2.18§,* 0.077

InvasivityInvasive (65) vs non-invasive (29) adenomas 1.66 ± 1.79 vs 1.00 ± 1.57 0.040Cavernous sinus invasion + (49) vs − (45) 1.74 ± 1.87 vs 1.00 ± 1.38° 0.035Sphenoidal sinus invasion + (17) vs − (77) 1.37 ± 1.15 vs 1.41 ± 1.65 0.515

Hormone secretion, CS (33) vs CNS (61) adenomas 2.06 ± 2.40 vs 1.13 ± 1.15°°° 0.092

Recurrences, recurrent (15) vs primary (79) adenomas 1.62 ± 1.59 vs 1.42 ± 1.78 0.516

Mean values were compared by Mann–Whitney univariate analysis ( + : Kruskall–Wallis for SSE) and significant P values areunderlined. The variability among subgroups was evaluated by F-test and significant P values are indicated by °P < 0.05,°°P < 0.01 and °°°P < 0.0001, and, for analysis according to SSE grade §P < 0.001 vs grade O, *P < 0.01 vs grade A/B.

Figure 4 LI in 94 untreated pituitary adenomas according to hormone secretion (means ± S.D.).

age (P < 0.01 in all cases) and to decrease with pre-operativetreatment (P < 0.025 in all cases).

CNS adenomas

No age-related LI variations was observed in this group (P =0.63, r = 0.06). LI was not significantly higher in invasivethan in non-invasive tumours (1.19 ± 1.19 vs 0.62 ± 0.73%,P = 0.11) but its variability was greater in the former group(P = 0.026). Similarly, invasion of either cavernous or sphe-noid sinus was not found to individually influence LI and nosignificant difference could be found according to patient’sgender, to the presence of visual field defects or to the gradeof SSE (data not shown). The mean LI was similar in nullcell and FSH/LH-secreting adenomas (0.95 ± 1.04 vs1.00 ± 1.06 respectively, P = NS).

www.endocrinology.org 107

Possible relationships between LI and regrowth of CNStumours were studied in two ways, leading to similar results:(i) considering prospective data only, LI was not significantlyhigher in recurrent than in primary tumours (1.45 ± 1.59 vs0.95 ± 0.96%, P = 0.32), although its variability was greaterin the former group (P = 0.008); (ii) including the analysis ofarchive material, paired analysis of data obtained in patientswho were re-operated on for tumour regrowth failed to dis-close significant differences between the two events(0.91 ± 1.05% for primary tumours vs 1.27 ± 1.52% for sub-sequent recurrences respectively, P = 0.48).

CS adenomas

Pre-operative treatment with either DA or SMS drugs wasfound to significantly reduce LI values (0.83 ± 1.04 vs

M L Jaffrain-Rea et al.: MIB-1 in pituitary tumours

(a) (b)

Figure 5 A case of huge GH/PRL-secreting adenomas observed in a 21-year-old female patient, operated on before and aftera 6 month pharmacological treatment with octreotide (OCT) and cabergoline. The LI decreased from 6% (a) to 2.7%(b) between the first and the second operation. LSABZ method. MoAb MIB-1, ×250.

2.15 ± 2.41% in treated vs in untreated tumours, P = 0.009)and its variability (P < 0.0001). An example of LI variationsafter pharmacological treatment with OCT and cabergolinein a young patient with a huge GH/PRL-secreting adenomais shown in Fig. 5. A significant negative linear correlationwas observed between LI values and the age of the patients(P = 0.025, r = 0.28).

Univariate analysis performed in untreated CS adenomas(n = 34) (Table 4) indicated that mean LI values were higherin macro- than in microadenomas (P = 0.013), and increasedwith the grade of ESS (P = 0.015) and invasiveness (P =0.048), including invasion of the cavernous sinus (P = 0.039).LI variability was also higher in macro- than in microaden-omas (P = 0.003) and in male than in female tumours (P =0.048).

According to a multivariate analysis taking into accountboth age and pre-operative treatment, LI was confirmed to be

Table 4 Factors influencing LI values in a series of 33 untreated CS pituitary adenomas. The number of cases is indicatedwithin brackets for each subgroup. SSE grade: suprasellar extension (Wilson 1984)

LI (%) Mann–Whitney(means ± S.D.) (+)

Gender, M (15) vs F (18) 2.88 ± 2.76 vs 1.38 ± 1.87° 0.048

Invasivity 2.86 ± 2.52 vs 1.22 ± 2.00° 0.048Invasive (16) vs non-invasive (13) 2.96 ± 2.65 vs 1.20 ± 1.88° 0.038Cavernous sinus invasion + (13) vs − (16) 0.43 ± 0.29 vs 1.99 ± 2.13 0.256Sphenoidal sinus invasion + (3) vs − (26)

Volume 0.54 ± 0.74 vs 2.47 ± 2.53°° 0.013Micro- (7) vs macroadenomas (22) 0.88 ± 0.94 vs 1.59 ± 2.26 vs 0.015SSE grade: O (10) vs A/B (17) vs C/D ( 2) 4.14 ± 2.73§,*

Mean values were compared by Mann–Whitney univariate analysis ( + : Kruskall–Wallis for SSE grade) and significant P valuesare underlined. The variability among subgroup was evaluated by F-test and the corresponding significant P values areindicated by °P < 0.05 and °°P < 0.005, and for SSE grade: §P = 0.001 vs O and *P = 0.009 vs A/B.

108 www.endocrinology.org

higher in macro- than in microadenomas (P = 0.006), toincrease with tumour invasiveness (P = 0.021), especially withinvasion of the cavernous sinus (P = 0.012), and with thedegree of SSE (P = 0.002). The difference between treated anduntreated patients remained significant when corrected for age,tumour volume and invasiveness (P < 0.015 in all cases),whereas the correlation with age disappeared, indicating age asa less important parameter in determining LI values.

Analysis of CS subgroups

In order to further analyse the influence of pre-operativepharmacological treatment in GH- and PRL-secreting aden-omas, each subgroup was evaluated separately by univariateanalysis (Table 5).

Pre-operative plasma GH, insulin-like growth factor-Iand PRL values were lower in acromegalic patients who

Endocrine-Related Cancer (2002) 9 103–113

Table 5 Influence of pre-operative pharmacological treatment on GH and/or PRL-secreting adenomas and distribution offunctional subgroups according to immunohistochemistry for pituitary hormones. The number of cases is indicated withinbrackets for each subgroup. ( + ) In GH-secreting adenomas, ‘others’ refer to treatment with DA drugs only or in associationwith OCT. For immunohistochemical classification, GH/glyP is for GH/glycoproteins and multiH for multihormonal adenomasrespectively

Pre-operative plasma hormone values Immunohistochemistry ( + )LI

(means ± S.E.M.) (ng/ml) (n)(means ±S.D.)

GH IGF-I PRL GH GH/glyP GH/PRL MultiH PRL (%)

GH-secreting adenomasUntreated (20) 32.3 ± 6.7 700.7 ± 61.1 22.5 ± 3.8 11 3 5 1 − 2.11 ± 2.16Treated (22) 10.6 ± 2.2°°° 443.5 ± 36.0°° 12.8 ± 3.2°° 10 4 7 1 − 1.20 ± 1.01°°,**OCT only (16) 6.9 ± 1.5°°° 418.9 ± 47.1°° 10.9 ± 3.3° 8 4 3 1 − 0.55 ± 0.98°°°,**Others( + )(6) 19.4 ± 4.5 543.3 ± 61.7 12.4 ± 9.8° 2 0 4 0 − 1.55 ± 1.02

PRL-secreting adenomasUntreated (6) − − 1665.5 ± 972.3 − − − 1 5 3.68 ± 3.30Treated (9) − − 205.2 ± 92.0° − − − 0 9 0.88 ± 1.20°,*

All mean values were compared by Mann–Whitney and the corresponding significant P values indicated by °P < 0.05 , °°P <0.01 and °°°P < 0.001 vs untreated cases. LI variability among subgroups was evaluated by F-test and the correspondingsignificant P values are indicated by *P < 0.05 and **P < 0.005.

received pre-operative pharmacological treatment comparedwith those who did not (P = 0.0007, P = 0.003 and P = 0.006respectively), including those treated by OCT alone(P < 0.0001, P = 0.004 and P = 0.018 respectively). Themean LI and LI variability were significantly higher inuntreated GH-secreting tumours compared with treatedtumours (P = 0.004 and P = 0.001 respectively), includingthose treated by OCT alone (P = 0.0009 and P = 0.003respectively).

Similarly, despite a marked variability, prolactinomapatients who received pre-operative treatment with DA drugshad lower pre-operative plasma PRL compared with thosewho did not (P = 0.01). Accordingly, mean LI values and LIvariability were lower in treated than in untreated pro-lactinomas (P = 0.025 and P = 0.026 respectively).

Correlations between histological featuresand MIB-1 index

Mitosis was observed in 8 of the 40 primary tumours studied(20%). The mean MI was extremely low (0.1 ± 0.2%), butstrongly correlated with LI by linear regression (P < 0.0001,r = 0.61). In contrast, no significant difference in either MIor LI could be found between tumours presenting with mild/absent atypia (n = 23) compared with those with moderate/marked atypia (n = 17).

Similar to data obtained in the whole series, the LI wassignificantly higher in untreated CS (3.40 ± 3.21%) comparedwith either treated CS (0.64 ± 0.81%, P = 0.0006) or CNSadenomas (0.97 ± 0.83, P = 0.001) in this subgroup oftumours. Mitosis could be detected only in CS tumours andthe MI was significantly higher in untreated CS comparedwith treated CS (0.15 ± 0.14 vs 0.05 ± 0.08%, P = 0.038).Moderate/marked nuclear atypias were present mostly in CS

www.endocrinology.org 109

tumours (60 vs 13.3%, P = 0.002) but were not significantlyinfluenced by pre-operative medical treatment.

A case of rapidly growing parasellarFSH-secreting tumour

A 33-year-old male patient was admitted at the NeurosurgeryUnit with a 6 month history of worsening headache withrecent diplopia and palpebral ptosis in his right eye, NMRimaging being strongly suggestive for a voluminous cavern-ous sinus meningioma. After incomplete fronto-temporal sur-gical resection, histological examination concluded that therewas an FSH-secreting pituitary tumour with marked cellularatypia and uncommonly elevated mitotic activity (mean LI =24%) (Fig. 6a). Three months later, the patient was re-admitted because of rapidly evolutive headache, vomiting,complete III° right nerve palsy and decreased vigilance.NMR imaging showed an impressive regrowth of the tumourwith hydrocephaly, and the patient died 10 days after re-operation because of neurological complications. No autopsycould be performed. Histologically, the tumour showed aneven more aggressive phenotype, with a mean LI exceeding50% (Fig. 6b). The diagnosis of probable pituitary carcinomawas finally retained on the basis of the rapidly fatal clinicalevolution and uncommon proliferative activity.

Discussion

Data presented herein confirm that, in agreement with pre-vious large series (Kitz et al. 1991, Thapar et al. 1996a, Mas-tronardi et al. 1999, Schreiber et al. 1999) most pituitarytumours have a low LI, with mean values usually less than3–4% and sporadic cases with higher proliferation rates.Careful analysis of this series and multivariate evaluation of

M L Jaffrain-Rea et al.: MIB-1 in pituitary tumours

(a) (b)

Figure 6 A case of highly aggressive FSH-secreting parasellar tumour observed in a 33-year-old male patient at first andsecond transfrontal surgery, performed over a 4 month span for rapid regrowth. (a) and (b) show areas of high LI in adishomogeneous tumour with a mean LI value of 24% at the first operation and exceeding 50% at the second. A pituitarycarcinoma? LSABZ method. MoAb MIB-1, ×400.

the data have helped to point out possible pitfalls in its inter-pretation and clarify some apparent discrepancies reported inthe literature.

According to most (Landolt 1987, Pegolo et al. 1995,Thapar et al. 1996a, Mastronardi et al. 1999) but not all(Zhao et al. 1999) authors, CS tumours have a higher pro-liferative potential than CNS tumours. In this study, the dif-ference between CS and CNS tumours appeared in untreatedcases, with significantly greater LI variations and a trendtowards higher LI values in the former subgroup. In contrast,because of the possible anti-proliferative effects of pre-operative pharmacological treatment, LI values in treated CSwere very similar to those observed in CNS. In fact, inaddition to the well-known anti-proliferative effects of dopa-minergic drugs on prolactinomas, possible in vivo anti-proliferative effects of OCT on somatotroph adenomas couldbe pointed out by this study and by a very recent specificreport (Losa et al. 2001). These data clearly indicate that theLI should not be interpreted only according to the functionalstatus of the tumour, but also to its pre-operative manage-ment. Interestingly, in agreement with previous findings(Pegolo et al. 1995), nuclear atypias were also more frequentin CS than in CNS adenomas, although there was no clearrelationship between mitosis and nuclear pleomorphism.Noteworthy, in CS adenomas, mitosis, but not atypias, werealso found to significantly decrease with pre-operative medi-cal treatment.

Another point of this study was the inverse relationshipobserved between the LI and the age of the patients, whichwas generally not reported previously (Buchfelder et al.1996, Blevins et al. 1998, Losa et al. 1998). In this study,this age-related LI distribution was limited to CS tumourswhereas, considering that CS and CNS adenomas were

110 www.endocrinology.org

almost equally represented, the older age of patients withCNS adenomas was likely to explain the age-related patternobserved on the whole series. This finding may be of clinicalimportance since the age-related pattern observed in CSadenomas is in keeping with their higher aggressivity inyoung patients (Melmed 1990, Abe et al. 1999), and similarfeatures have been recently proposed for CNS adenomas(Losa et al. 2000b). However, data from multivariate analysisindicate age as a secondary determinant factor when com-pared with pre-operative treatment or macroscopic featuresof the tumour itself.

Most (Landolt et al. 1987, Knosp et al. 1991, Thapar etal. 1996a, Mastronardi et al. 1999, Zhao et al. 1999) but notall (Gandour-Edwards et al. 1995, Kawamoto et al. 1995,Pegolo et al. 1995, Yonezawa et al. 1997, Losa et al. 2000b)studies have reported a higher proliferation rate in invasivethan non-invasive tumours. Discrepancies may in part reflectdifferences in the definition of invasiveness, since about 40%of pituitary adenomas show gross invasion of the surround-ing structures, contrasting with 70–95% showing histologicalevidence for dural infiltration, depending on tumour volumeand functional characteristics (Scheithauer et al. 1986,Selman et al. 1986). Histologically proved dural invasion hasbeen associated with higher LI values (Knosp et al. 1989,Kitz et al. 1991). In our series, the overall percentage ofinvasion was about 65%, which, with a few exceptions(Pegolo et al. 1995), was higher than commonly reported.This might be explained by the definition retained for thestudy — localised macroscopic dural infiltration was consid-ered significant — and by the high percentage of macroad-enomas referred to our Neurosurgery Units (Fraioli et al.1995). Invasiveness was confirmed by multivariate analysisto be associated with a higher LI in the whole series and in

Endocrine-Related Cancer (2002) 9 103–113

CS adenomas. Supporting some (Knosp et al. 1991, Mastron-ardi et al. 1999) but not all (Kawamoto et al. 1995) previousdata, invasion of the cavernous sinus, but not of the sphenoidsinus (Gandour-Edwards et al. 1995), was associated with ahigher LI in most cases, although significance was notreached for CNS tumours, which only displayed a highervariability in LI values. This finding supports the hypothesisthat peculiar biological mechanisms may lead to invasion ofthe different perisellar tissues by pituitary tumours(Jaffrain-Rea et al. 1998, Kawamoto et al. 1996a,b).

Most studies agree on the lack of correlation betweentumour volume and proliferative activity in pituitary aden-omas (Pegolo et al. 1995, Yonezawa et al. 1997). However,surgical indications for microadenomas are almost limited toCS tumours, so that ignoring the functional status of theadenomas may introduce a significant bias and mask volume-related variations. In the present study, only LI variabilitywas found to increase with tumour volume on the wholeseries. However, mean LI values and its variability weremarkedly higher in macro- than in micro-CS adenomas, afinding which was further confirmed by multifactorial analy-sis. In addition, mean LI values and variability increased withSSE in this group. Higher LI values have also been recentlyreported in macro- vs micro-ACTH-secreting adenomas(Losa et al. 2000a) and in macro- vs micro-PRL-secretingadenomas (Turner et al. 2000). The high LI values observedherein in untreated prolactinomas, which were all macroad-enomas, contrasting with the low values reported in poorlyevolutive microprolactinomas (Nishioka et al. 2001) furtherreinforce this point. In this series, this finding could beextended to GH-secreting adenomas (data not shown). Thisis in keeping with the slowly growing potential of most CSmicroadenomas in clinical practice, while further supportingthe frequent aggressivity of CS macroadenomas. In contrast,no significant LI variations was found in CNS adenomasaccording to their volume. This finding is in agreement witha recent report (Losa et al. 2000b) and further supports theslowly growing potential displayed by most CNS adenomas.

Despite some exceptions (Katoh et al. 1995), LI is cur-rently viewed as a reliable marker of aggressivity for endo-crine neoplasms (Vargas et al. 1997a,b, Clarke et al. 1998).Some authors have suggested that regrowing pituitarytumours have a higher LI than non-regrowing tumours(Shibuya et al. 1992, Ekramullah et al. 1996, Abe et al. 1997,Mizoue et al. 1997). Again, we recommend that the possibleprognostic value of LI should be regarded differently in CSand in CNS tumours. In fact, although high LI values aremore commonly seen in untreated CS tumours, most of themare GH and/or PRL-secreting tumours which will be respon-sive to post-operative pharmacological treatment, so that theLI is expected to be mainly useful for the management ofCNS or CS resistant to pharmacological therapy. Our datasupport recent evidence that primary and recurrent CNStumours do not significantly differ (Yonezawa et al. 1997),

www.endocrinology.org 111

and that LI alone seems to poorly predict the risk of post-operative regrowth in this group (Losa et al. 2000b). Thismay be explained by factors other than cell proliferationbeing involved in the kinetics of tumour growth (i.e. tumourvascularisation and apoptotic index) and by the quality ofsurgical resection as a key prognostic factor. Nonetheless,since a correlation may exist between in vitro (Atkin et al.1997) or in vivo (Hsu et al. 1993, Ekramullah et al. 1996)tumour doubling time, we believe that LI values should bekept in mind for an adequate follow-up of post-operativeresidual tumours.

Special attention should be paid to sporadic casespresenting with unusually high LIs. Despite the fact that dataobtained on some endocrine neoplasms suggest that LIthresholds can be proposed which can help disclosing specialaggressivity or malignancy (Vargas et al. 1997a,b, Clarke etal. 1998), evidence for metastasis is currently recognised asthe sole criteria for the definition of pituitary carcinomas(Pernicone et al. 1997). According to this definition, thesetumours are extremely rare. It should be remembered thatpituitary carcinomas are believed to arise from benign pro-liferations secondary to de novo molecular events (Karga etal. 1992, Thapar et al. 1996b), an evolution which may takemore than 10–15 years. Thus, despite high LI values reportedin most documented pituitary carcinomas (Thapar et al.1996a),‘paradoxically’ low values can be observed at firstsurgery (Pernicone et al. 1997). We should also point outthat very high LI values in other tumours are typically associ-ated with malignancy, regardless of the presence of metasta-sis (Katoh et al. 1995, Garcia et al. 1997). Thus, as illustratedin the brief case report presented herein, current criteria maysometimes be too restrictive for clinical purposes, and veryhigh LI values may help identify a subset of pituitarytumours with a highly aggressive, locally ‘malignant’ or life-threatening behaviour prevailing on their ability to metastas-ise. As a general rule, accepting the 90° percentile as areasonable threshold to disclose ‘unusual LI values’, mostpituitary adenomas with LI > 3% should be regarded withspecial attention. On the basis of the present study, we sug-gest to adapt this cut-off to the functional characteristics ofthe tumour, i.e. 2.5% for CNS adenomas or treated CS, and6.0% for untreated CS. This may help the pathologist to dis-close highly aggressive tumours and the physician to plan amore adequate short-term follow-up and, if necessary, a moreaggressive therapeutic schedule (Knosp et al. 1995, Kaltsaset al. 1998).

In conclusion, we believe that immunostaining for theMIB-1 LI should be part of the routine examination of oper-ated pituitary tumours, provided that both pathologists andphysicians keep aware of the functional status of the tumourand the presence of any pre-operative pharmacological treat-ment. In fact, despite its value for clinical decision makingbeing restricted to a subset of tumours, it may represent aprecious tool for the diagnosis and management of tumours

M L Jaffrain-Rea et al.: MIB-1 in pituitary tumours

displaying an unusually aggressive phenotype and/or phar-macological resistance to conventional medical therapy. Onthe other hand, pitfalls in its interpretation should be kept inmind by researchers interested in pituitary tumour patho-genesis and in disclosing factors associated with an increasedkinetic activity.

Acknowledgements

We are indebted to Remo Pompei for his excellent technicalassistance. The present study was partially supported bygrants from MURST and AIRC.

References

Abe T, Sanno N, Osamura YR & Matsumoto K 1997 Proliferativepotential in pituitary adenomas: measurement by monoclonalantibody MIB-1. Acta Neurochirirurgica 139 613–618.

Abe T, Tara LA & Ludecke DK 1999 Growth-hormone-secretingpituitary adenomas in childhood and adolescence: features andresults of transnasal surgery. Neurosurgery 45 1–10.

Atkin SL, Green VL, Hipkin LJ, Landolt AM, Foy PM, JeffreysRV & White MC 1997 A comparison of proliferation indices inhuman anterior pituitary adenomas using formalin-fixed tissueand in vitro cell culture. Journal of Neurosurgery 87 85–88.

Bevan JS, Webster J, Burke CW & Scanlon MF 1992 Dopamineagonist and pituitary tumor shrinkage. Endocrine Reviews 13220–240.

Blevins LS Jr, Verity DK & Allen G 1998 Aggressive pituitarytumors. Oncology 12 1307–1312.

Buchfelder M, Fahlbusch R, Adams EF, Kiesewetter F & ThieraufP 1996 Proliferation parameters for pituitary adenomas. ActaNeurochirurgica. Supplement 65 18–21.

Cattoretti G, Becker MH, Key G, Duchrow M, Schluter C, GalleJ & Gerdes J 1992 Monoclonal antibodies against recombinantparts of the Ki-67 antigen (MIB1 and MIB3) detect proliferatingcells in microwave-processed formalin-fixed paraffin sections.Journal of Pathology 168 357–363.

Clarke MR, Weyant RJ, Watson CG & Carty SE 1998 Prognosticmarkers in pheochromocytoma. Human Pathology 29 522–526.

Ekramullah SM, Saitoh Y, Ohnishi T, Arita N, Taki T &Hayakawa T 1995 Effects of bromocriptine on staining indicesKi-67, proliferating cell nuclear antigen and nucleolar organizerregion number in pituitary adenomas. NeurologicaMedicochirurgica 35 221–226.

Ekramullah SM, Saitoh Y, Arita N, Ohnishi T & Hayakawa T1996 The correlation of Ki-67 staining indices and tumordoubling time in regrowing non-functioning pituitary adenomas.Acta Neurochirurgica 138 1449–1455.

Fraioli B, Esposito V, Santoro A, Iannetti G, Giuffre R & CantoreG 1995 Transmaxillosphenoidal approach to tumors invading themedial compartment of the cavernous sinus. Journal ofNeurosurgery 82 63–69.

Gandour-Edwards R, Kapadia SB, Janecka IP, Martinez AJ &Barnes L 1995 Biologic markers of invasive pituitary adenomasinvolving the sphenoid sinus. Modern Pathology 8 160–164.

Garcia R, Bueno A, Castanon S, Ruiz-Barnes P, Maria de CamposJ, Kusak E, Fortes JR, Ortiz F & Sarasa JL 1997 Study of theDNA content by flow cytometry and proliferation in 281 braintumors. Oncology 54 112–117.

112 www.endocrinology.org

Hsu DW, Hakim F, Biller BM, de la Monte S, Zervas NT,Kilibanski A & Hedley-White ET 1993 Significance ofproliferating cell nuclear antigen index in predicting pituitaryadenoma recurrence. Journal of Neurosurgery 78 753–761.

Jaffrain-Rea ML, Derome P, Bataini JP, Thomopoulos P, BertagnaX & Luton JP 1993 Influence of radiotherapy on long-termrelapse in clinically non-secreting pituitary adenomas. Aretrospective study (1970–1988). European Journal of Medicine2 398–403.

Jaffrain-Rea ML, Petrangeli E, Lubrano C, Minniti G, Di StefanoD, Sciarra F, Frati L, Tamburrano G, Cantore G & Gulino A1998 Epidermal growth factor binding sites in human pituitarymacroadenomas. Journal of Endocrinology 158 425–433.

Jaffrain-Rea ML, Ferretti E, Toniato E, Cannita K, Santoro A, DiStefano D, Ricevuto E, Maroder M, Tamburrano G, Cantore G,Gulino A & Martinotti S 1999 p16 (INK4a, MTS-1) genepolymorphism and methylation status in human pituitarytumours. Clinical Endocrinology 51 317–325.

Kaltsas GA, Mukherjee JJ, Plowman PN, Monson JP, GrossmanAB & Besser GM 1998 The role of cytotoxic chemotherapy inthe management of aggressive and malignant pituitary tumors.Journal of Clinical Endocrinology and Metabolism 83 4233–4238.

Karga HJ, Alexander JM, Hadley-White ET, Klibanski A &Jameson JL 1992 Ras mutations in human pituitary tumors.Journal of Clinical Endocrinology and Metabolism 74 914–919.

Katoh R, Bray CE, Suzuki K, Komiyama A, Hemmi A, Kawaoi A,Oyama T, Sugai T & Sasou S 1995 Growth activity inhyperplastic and neoplastic human thyroid determined by animmunohistochemical staining procedure using monoclonalantibody MIB-1. Human Pathology 26 139–146.

Kawamoto H, Uozumi T, Kawamoto K, Arita K, Yano T &Hiroata T 1995 Analysis of the growth rate and cavernous sinusinvasion of pituitary adenomas. Acta Neurochirurgica 136 37–43.

Kawamoto H, Uozumi T, Kawamoto K, Arita K & Hiroata T1996a Type IV collagenase activity and cavernous sinusinvasion in human pituitary adenomas. Acta Neurochirurgica138 390–395.

Kawamoto H, Kawamoto K, Mizoue T, Uozomi T, Arita K &Kurisu K 1996b Matrix metalloproteinase-9 secretion by humanpituitary adenomas detected by cell immunoblot analysis. ActaNeurochirurgica 138 1442–1448.

Kitz K, Knosp E, Koos WT & Korn A 1991 Proliferation inpituitary adenomas: measurement by MAb KI 67. ActaNeurochirurgica. Supplement 53 60–64.

Knosp E, Kitz K & Perneczky A 1989 Proliferation activity inpituitary adenomas: measurement by monoclonal antibody Ki-67.Neurosurgery 25 927–930.

Knosp E, Kitz K, Steiner E & Matula C 1991 Pituitary adenomaswith parasellar invasion. Acta Neurochirurgica. Supplement 5365–71.

Knosp E, Perneczky A, Kitz K, Grunert P & Wild A 1995 Theneed for adjunctive focused radiotherapy in pituitary adenomas.Acta Neurochirurgica. Supplement 63 81–84.

Lamberts SWJ, Van der Lely AJ, De Herder WW & Hofland LJ1996 Drug therapy: octreotide. New England Journal ofMedicine 4 246–254.

Landolt AM, Shibata T & Kleihues P 1987 Growth rate of humanpituitary adenomas. Journal of Neurosurgery 67 803–806.

Losa M, Franzin A, Mortini P, Terreni MR, Mangili F &Giovannelli M 1998 Usefulness of markers of cell proliferation

Endocrine-Related Cancer (2002) 9 103–113

in the management of pituitary adenomas. Clinical Science 95129–135.

Losa M, Barzaghi RL, Mortini P, Franzin A, Mangili F, TerreniMR & Giovanelli M 2000a Determination of the proliferationand apoptotic index in adrenocorticotropin-secreting pituitarytumors: comparison between micro- and macroadenomas.American Journal of Pathology 156 245–251.

Losa M, Franzin A, Mangili F, Terreni MR, Barzaghi R, Veglia F,Mortini P & Giovanelli M 2000b Proliferation index innon-functioning pituitary adenomas: clinical characteristics andlong-term follow-up. Neurosurgery 47 1313–1318.

Losa M, Ciccarelli E, Mortini P, Barzaghi R, Gaia D, Faccani G,Papotti M, Mangili F, Terreni MR, Cavnanni F & Giovanelli M2001 Effects of octreotide treatment on the proliferation andapoptotic index of GH-secreting pituitary adenomas. Journal ofClinical Endocrinology and Metabolism 86 5194–5200.

Mastronardi L, Guiducci A, Spera C, Puzzilli F, Liberati F &Maira G 1999 Ki-67 labelling index and invasiveness amonganterior pituitary adenomas: analysis of 103 cases using theMIB-1 monoclonal antibody. Journal of Clinical Pathology 52107–111.

Melmed S 1990 Acromegaly. New England Journal of Medicine322 966–977.

Melmed S 1999 Pathogenesis of pituitary tumors. Endocrinologyand Metabolism Clinics of North America 28 1–12.

Mizoue T, Kawamoto H, Arita K, Kurisu K, Tominaga A &Uozomi T 1997 MIB1 immunopositivity is associated with rapidregrowth of pituitary adenomas. Acta Neurochirurgica 139 426–432.

Nishi T, Goto T, Takeshima H, Hamada K, Tada K, Saito Y,Kochi M, Kuratsu J & Ushio Y 1999 Tissue factor expressed inpituitary adenoma cells contributes to the development ofvascular events in pituitary adenomas. Cancer 86 1354–1361.

Nishioka H, Ito H, Haraoka J & Hirano A 2001 Growth potentialof human prolactinomas. Surgical Neurology 55 213–217.

Pegolo G, Buckwalter JG, Weiss MH & Hinton DR 1995 Pituitaryadenoma. Correlation of the cytologic appearance with biologicbehavior. Acta Cytologica 39 887–892.

Pernicone PJ, Scheithauer BW, Sebo TJ, Kovacs KT, Horvath E,Young WF Jr, Lloyd RV, Davis DH, Guthrie BL & SchoeneWC 1997 Pituitary carcinoma: a clinicopathologic study of 15cases. Cancer 79 804–812.

Plowman PN 1995 Radiotherapy for pituitary tumors. Baillere’sClinical Endocrinology and Metabolism 9 407–420.

Scheithauer BW, Kovacs KT, Laws ER Jr & Randall RV 1986Pathology of invasive pituitary tumors with special reference tofunctional classification. Journal of Neurosurgery 65 733–744.

Schreiber S, Saeger W & Ludecke DK 1999 Proliferation markersin different types of clinically non-secreting pituitary adenomas.Pituitary 1 213–220.

Selman WR, Laws ER Jr, Scheithauer BW & Carpenter SM 1986The occurrence of dural invasion in pituitary adenomas. Journalof Neurosurgery 64 402–407.

Shibuya M, Saito F, Miwa T, Davis RL, Wilson CB & Hoshino T1992 Histochemical study of pituitary adenomas with Ki-67 and

www.endocrinology.org 113

anti-DNA polymerase alpha monoclonal antibodies,bromodeoxyuridine labeling, and nucleolar organizer regioncounts. Acta Neuropathologica 84 178–183.

Shimon I, Hinton DR, Weiss MH & Melmed S 1998Prolactinomas express human heparin-binding secretorytransforming gene (hst) protein product: marker of tumourinvasiveness. Clinical Endocrinology 48 23–29.

Tampanaru-Sarmesiu A, Stefaneau L, Thapar K, Kontogeorgos G,Sumi T & Kovacs K 1998 Transferrin and transferrin receptor inhuman hypophysis and pituitary adenomas. American Journal ofPathology 152 413–422.

Thapar K, Kovacs KT, Stefaneanu L, Scheithauer BW, Horvath E,Lloyd RV, Li J & Laws ER Jr 1996 Proliferative activity andinvasiveness among pituitary adenomas and carcinomas: ananalysis using the MIB-1 antibody. Neurosurgery 38 99–107.

Thapar K, Scheithauer BW, Kovacs K, Pernicone PJ & Laws ERJr 1996 p53 expression in pituitary adenomas and carcinomas:correlation with invasiveness and tumor growth fractions.Neurosurgery 38 765–771.

Thapar K, Kovacs K, Stefaneanu L, Scheithauer B, Killinger DW,Lioyd RV, Smyth HS, Barr A, Thorner MO, Gaylinn B & LawsER Jr 1997 Antiproliferative effect of the somatostatin analogueoctreotide on growth hormone-producing pituitary tumors:results of multicenter randomized trial. Mayo Clinic Proceedings72 893–900.

Thapar K, Kovacs K, Stefaneanu L, Scheithauer B, Killinger DW,Lioyd RV, Smyth HS, Barr A, Thorner MO, Gaylinn B & LawsER Jr 1997 Overexpression of the growth-hormone-releasinghormone gene in acromegaly-associated pituitary tumors. Anevent associated with neoplastic progression and aggressivebehavior. American Journal of Pathology 151 769–784.

Turner HE & Wass JAH 1999 Are markers of proliferationvaluable in the histological assessment of pituitary tumors?Pituitary 1 147–151.

Turner HE, Nagy Z, Gatter KC, Esiri MM, Wass JA & Harris AL2000 Proliferation, bcl-2 expression and angiogenesis in pituitaryadenomas: relationship to tumour behaviour. British Journal ofCancer 82 1441–1445.

Vargas MP, Vargas HI, Kleiner DE & Merino MJ 1997a The roleof prognostic markers (Mib-1, RB and bcl-2) in the diagnosis ofparathyroid tumors. Modern Pathology 10 12–17.

Vargas MP, Vargas HI, Kleiner DE & Merino MJ 1997bAdrenocortical neoplasms: role of prognostic markers MIB-1,p53 and RB. American Journal of Surgical Pathology 21 556–562.

Wilson CB 1984 A decade of pituitary microsurgery: The HerbertOlivecrona Lecture. Journal of Neurosurgery 61 814–833.

Yokoyama S, Thapar K, Kovacs K & Stefaneanu L 1998Localization of insulin-like growth factor-II mRNA in humanpituitary adenomas. Virchow’s Archives 432 241–246.

Yonezawa K, Tamaki N & Kokunai T 1997 Clinical features andgrowth fractions of pituitary adenomas. Surgical Neurology 48494–500.

Zhao D, Tomono Y & Nose T 1999 Expression of p27kip1 andKi-67 in pituitary adenomas: an investigation of marker ofadenoma invasiveness. Acta Neurochirurgica 141 187–192.