CD44, intercellular adhesion molecule-1 and vascular cell adhesion molecule-1: biomarkers in search of validation in lymphomas

In this issue of Leukemia and Lymphoma, Shah and colleagues report on the ability of three exploratory biomarkers to predict the outcomes of patients with indolent lymphoma [1]. The investigators found that three proteins that mediate the interaction between lymphoma cells and the microenvironment (CD44, intercellular adhesion molecule-1 [ICAM-1] and vascular cell adhesion molecule-1 [VCAM-1]) were associated with progression-free survival and overall survival (although only ICAM-1 maintained its predictive value in a multivariate analysis).

The Follicular Lymphoma International Prognostic Index (FLIPI) is currently the best prognostic tool available [2]. It is easy to use and is valid in the front-line setting and the relapsed setting in the pre- and post-rituximab eras [3,4]. Oddly, in this era of molecular medicine and targeted agents, the FLIPI uses very little information that is tumor specific – only lactate dehydrogenase (LDH) is reflective of the actual tumor biology. The FLIPI2, a prospectively developed prognostic tool meant to replace the FLIPI, incorporates β₂-microglobulin rather than LDH, but again, clinical factors representing tumor bulk and co-morbid status dominate the calculation [5]. The absence of biology-specific biomarkers is not for the absence of trying. Over the past decade, investigators have published positive data on dozens of exploratory biomarkers. Some of them, such as proliferation index [6] or CD68 [7], are widely available, while others, such as gene expression profiling [8], are restricted to research laboratories. Given the profusion of published data, two questions arise. First, why do none of these biomarkers make it into clinical practice? And second, why do we continue to study biomarkers that will most probably never be used? The study by Shah and colleagues provides hints to both questions.

Compared to many biomarker studies, the study reported by Shah has several strengths. It involved roughly 100 patients studied prospectively and treated uniformly. The median duration of follow-up was over 15 years, making it uniquely qualified to comment on overall survival. This strength, however, also posed some problems. Standard treatment two decades ago did not involve rituximab. Moreover, the chemotherapy regimens that were used would rarely be seen today. The reported median progression-free survival was remarkably long, but the definition of progression differed from today's standards. The serum biomarker testing, although relatively inexpensive and theoretically widely available, is not standardized between institutions, and there is no accepted definition of high versus low serum levels. Finally, after multivariate analysis with adjustment for the FLIPI, only ICAM-1 maintained predictive value, suggesting that CD44 and VCAM-1 levels could be largely related to tumor bulk or LDH. All of these are examples of how challenging it can be to validate a biomarker. Studies that use small sample size or retrospective data collection are prone to bias or statistical chance, while larger studies rarely include a homogeneous population. Studies that have limited follow-up may look promising initially but are later shown to be false, while studies with longer follow-up are subject to changing clinical patterns that render the results obsolescent. For example, tumor-associated macrophages (TAMs) were associated with a poor prognosis in the pre-rituximab era, but lost their predictive value following the introduction of rituximab. The only remedy to these problems, as Shah points out, is to validate biomarkers in the context of larger studies with more contemporary therapies. As clinical databases and tissue banks become more common in academic institutions and the cooperative groups, we can hope that investigators will pool resources to evaluate some of the more promising biomarkers on a larger scale.

Even in the absence of confirmatory studies, exploratory biomarker studies can serve an important purpose. Taken as a whole, they inform our understanding of disease biology. A single study using enzyme-linked immunosorbent assay (ELISA) to evaluate serum ICAM-1 levels suggests a role for the microenvironment, but may be discounted due to methodological issues. Combined with immunohistochemistry for CD68/TAM and gene expression profiling of the host immune response, however, it becomes difficult to imagine that the microenvironment is not a vital factor in lymphoma biology. In another example, the often discussed but rarely clinically useful immunoglobulin variable heavy chain gene (IgVH) mutation status suggests a role for signaling through the B-cell receptor (BCR) in CLL pathogenesis. Combined with knowledge of biased V gene usage and ZAP-70, however, there is a further rationale for targeting the BCR signaling cascade with therapeutic agents now in development. Indeed, there remains a role for exploratory biomarker studies, even if most of them never attain clinical utility. In the coming years we will see an increased push to develop predictive biomarkers (those that identify subgroups of patients most likely to experience a given outcome in response to a specific therapy) as opposed to prognostic biomarkers (those that provide information about a disease regardless of treatment). We should keep in mind that all biologically sound prognostic biomarkers have the potential to become predictive biomarkers pending the development of the right therapies.

Potential conflict of interest:

Disclosure forms provided by the authors are available with the full text of this article at www.informahealthcare.com/lal.