Ki67 and PIM1 expression predict outcome in mantle cell lymphoma treated with high dose therapy, stem cell transplantation and rituximab: a Cancer and Leukemia Group B 59909 correlative science study

Introduction

Mantle cell lymphoma (MCL) is considered incurable with a median overall survival of 3–4 years using conventional chemotherapy [1-4]. MCL is characterised clinically by a pattern of continuous relapse following conventional chemotherapy. New therapies are needed and recent studies using high dose chemotherapy regimens with or without stem cell transplant as first line treatment have suggested benefit in terms of response and remission duration; however, whether overall survival is improved upon compared with conventional therapies remains to be proven [3,5-9].

Pathologically, MCL is composed of small, mature appearing lymphocytes with slight nuclear irregularities. The nodal infiltration is usually diffuse, although nodular and, rarely, mantle zone variants may be seen. The relatively aggressive nature is characterised by the frequent presence of mitotic figures, a feature unusual in other B-cell lymphomas of small lymphocytes. At the molecular genetic level, the vast majority of cases harbour an IGH-CCND1 fusion as a consequence of the t(11;14)(q13;q32) resulting in overexpression of cyclin D1 protein, a diagnostic hallmark of this lymphoma. Only rare cases lack this abnormality and likely have abnormalities in other cyclins, such as cyclin D2 [10].

To elucidate molecular features of MCL, gene expression profiling has been performed in MCL and a recent seminal study has shown that a quantitative proliferation gene signature is a powerful predictor of outcome in MCL [11]. Several groups have investigated the significance of proliferation in MCL as measured by single markers such as Ki67 and have shown that high proliferative fraction was associated with poor outcome in the studied populations [12,13]. Indeed, other indicators of increased proliferation such as mitotic counts and blastoid morphology have been associated with poor prognosis in MCL [1,14]. These studies have been retrospective in nature and were performed on heterogeneously treated patients. Furthermore, prognostic significance of the proliferation index has not been examined in the context of combined high-dose therapy and rituximab.

Blastoid MCL is a variant of MCL with an extremely poor prognosis in which PIM1 has been found to be over-expressed [13,15,16]. PIM1, a serine/threonine kinase and protooncogene, has been implicated in lymphomagenesis with proliferative and anti-apoptotic effects [17-21]. It is a member of a family of proteins containing homologues PIM2 and PIM3, and is transcriptionally regulated by cytokines, mitogens and numerous growth factors [22-24]. The enzymatic substrates reported for PIM kinases are diverse (p100, Bad, NFATc1, p21/Cip1, Cdc25a, SOCS), suggesting that PIM kinases play a central regulatory role in numerous processes [25-30]. Elevated expression of PIM kinases has been described predominantly in murine and human leukemias and lymphomas [24,31-33]. Studies using either PIM transgenic or PIM knockout mice suggest that overlapping functions exist for PIM1 and PIM2, and that the full transforming potential of PIM kinases cannot be realised in vivo without cooperation from other oncogenes, including MYC and MYCN [33,34]. PIM1 has been implicated in the development of leukemias, lymphomas, in late progression events and also in prostate cancer [35]. Therefore, PIM1 is a target of drug development programs. With the promise of improved outcome in MCL patients treated with high dose therapy and rituximab, we have examined the prognostic importance of the proliferation marker Ki67 and PIM1 in predicting outcome in patients enrolled in Cancer and Leukemia Group B (CALGB) 59909 [36].

Methods

Results

Discussion

The clinical outcome for MCL is poor; however, substantial variability exists with some patients succumbing in less than a year while others may survive 10 years or longer [1,4,38,39]. Since its recognition as a distinct entity, numerous studies have been published identifying prognostic factors that have ranged from standard morphologic features such as blastoid morphology or mitotic counts [14,40] to increased Ki67 index [12,13,41], P53 overexpression and other cell cycle protein expression abnormalities [42-44]. More sophisticated comparative genomic hybridisation abnormalities or gene expression profiles have also been shown to have prognostic value [11,45,46]. The importance of cell cycle and proliferation in determining the clinical behaviour of MCL has been confirmed by expression profiling and a quantitative proliferation signature has been proposed based on 20 genes that belonged to a larger set of proliferation signature genes that were highly variably expressed in MCL and associated with survival [11]. Although these latter developments are exciting and potentially useful, they are still impractical for routine use.

Most of the studies to date using conventional techniques such as IHC, have confirmed the prognostic value of Ki67 staining and, in fact, its use has been advocated in the new MCL international prognostic index [47]. Most studies evaluating this proliferation marker have been performed outside the context of a single clinical trial [12,13,41,48,49]. Investigators have also found Ki67 level to remain predictive of outcome in multivariate analysis [12,13,41,48]. However, patient populations were clinically heterogeneous and manual methods were used for evaluation of Ki67 (Table III).

A prognostic marker must be evaluated in the context of the treatment regimen of the study group, as novel treatment approaches may alter the utility of prognostic markers. Rituximab is now commonly used in MCL treatment. A single study has evaluated Ki67 in patients treated with rituximab-CHOP chemotherapy and reported that high Ki67 (≥ 30%) predicted poor outcome [50]. We have now demonstrated that high Ki67 remains a poor prognostic indicator in the setting of high dose chemotherapy/stem cell transplantation regimens containing rituximab. Of note, our cutoff of 35% is similar to the high level suggested by Determann [50]. To our knowledge, the current study is also the first to use IA to more objectively quantify Ki67 in MCL. Indeed, computerised methods of Ki67 quantitation have been shown to be reproducible within and between operators, with correlation coefficients over 0.9 [51,52], a prerequisite for widespread use outside the setting of controlled clinical trials. Indeed, a recent study showed problems with reproducible scoring of Ki67 when a semiquantitative visual estimate was used [53]. Thus use of IA is suggested in future practice. It is interesting that the proliferation signature defined by gene expression analysis did not include Ki67 [11]. This may be due to differences in protein and RNA expression or complex interactions of multiple genes serving to outweigh Ki67 as part of the model. Nevertheless, Ki67 as a single biomarker, correlated with proliferation signature in the small number of cases in which both were available [11]. It appears to be a robust MCL prognostic marker, both in conventionally treated patients and in patients receiving high dose therapy with rituximab.

Blastoid MCL is an aggressive variant of MCL and PIM1 has been found to be overexpressed in this variant, leading some to propose therapeutic targeting of this molecule [15,16,54]. We, therefore, also examined expression of PIM1 in these cases. PIM1 has been implicated in anti-apoptosis through the phosphorylation of p21^cip1/waf1 [55], which is thought to mediate survival by binding to and preventing the activation of pro-caspase-3 [56]. Most recently, the PIM1 kinase has been found to promote inactivation of the pro-apoptotic Bad protein by phosphorylating it on the Ser¹¹² gatekeeper site [57]. PIM1 is a target of the gp130-mediated STAT3 signalling pathway [58]. It is stabilised by HSP-90 [59] and it phosphorylates the tyrosine phosphatase Cdc25 [60], which plays a critical role in G1/S cell cycle transition. In addition, PIM1 and PIM2 regulate rapamycin-resistant T cell growth and survival [61]. Thus, PIM kinases have defined biological roles in cell survival, proliferation and differentiation [24]. Consistent with these roles, PIM1 expression appears to be a marker of aggressive disease in MCL, even in the absence of blastoid morphology. Indeed, PIM1 was scored as > 1+ in 48% of cases and there was no association with blastoid morphology or Ki67.

Since blastoid histology is known to be associated with poor outcome, it may be that PIM1 expression also identifies cases with blastoid-like biologic features while not demonstrating overt “blastoid” cytologic features. Because of the small number of cases available for this study, additional study of PIM1 and other PIM kinases is warranted. The kinase activity of PIM1 is directly linked to the expression level of PIM1. But mutations in PIM1 have been reported in diffuse large B-cell lymphomas that result in altered function [62]. Perhaps measuring the kinase activity of PIM1 or phosphorylated substrates of PIM1 by IHC (as a surrogate of PIM1 activity) may be even more informative than the protein expression level.

New therapies are needed for MCL. The poor outcome and lack of standardised therapy has spurred numerous clinical trials. Aggressive treatment protocols with or without autologous stem cell transplantation have been attempted with promising results, including a relatively high rate of durable remissions [5,6,63,64]. CALGB 59909 follows in a similar vein [36]. As in prior studies with conventional therapies, Ki67 staining pattern inversely correlated with EFS and thus appears to remain an important biologic marker of outcome. Taken together, the evidence is now overwhelming in support of the Ki67 proliferative index as a prognostic marker in MCL. As shown in this study, this includes those patients treated with high dose therapies, stem cell transplantation and rituximab. Issues of technical standardisation remain to be completely resolved; however, Ki67 is a robust marker with control tissues commonly available such as hyperplastic lymph nodes or tonsil that can easily be evaluated for quality of stain and IA should make reporting more reproducible. In combination with IA, we believe it is time to develop prognostic models that incorporate this widely available biologic marker. In particular, we agree with the proposal to include the Ki67 index in prognostic scoring systems [47] with the goal of increasing our ability to stratify patients at diagnosis and select appropriate therapy.