Section 1 -

Borderline Tumors: A Small Step Forward Towards a Rational Tumor Classification Juan Rosai Milano, Italy

One of the most revered paradigms in pathology is the distinction of tumors into two categories: benign and malignant. The concept, which goes back to antiquity, was clearly enunciated by Gabriele Fallopio (of the fallopian tubes fame) in the 1500's, when he stated that tumors can be subdivided "in benignos et malignos", and given an authoritative stamp of approval by Rudolph Virchow in his seminal book "Cellular Pathology" (1871) through his distinction of Geschwülste (tumors) into "gutartige" (benign) and "bösartige" (malignant). Indeed, one of the most appreciated qualities of the surgical pathologist is his or her ability in making this distinction Arthur Purdy Stout himself confided, towards the end of this career, that the aspect of surgical pathology that was closest to his heart was how to tell "a benign from a malignant tumor so that the surgeon can proceed on the best form of treatment". The question to be explored in this presentation is whether this manicheist approach is biologically accurate and clinically sound. We believe at present that neoplasia is a genetic disease resulting from germ line and/or somatic mutations, and that several such mutations are required for the fully developed neoplastic phenotype to be established. This implies a step-wise process, best exemplified by the famous diagram proposed by Bert Vogelstein for the development of colorectal adenocarcinoma (the so-called "Vogelgram"), subsequently adopted with modifications and some degree of imagination for many other sites. This model portrays a step-wise (although not necessarily orderly) progression during the process of carcinogenesis, a scheme which had actually been proposed many years before by pathologists on the basis of purely morphologic observations. It has also become apparent that the two-valued approach that divides tumors sharply into benign and malignant types, besides failing to accurately reflect the tumor's biology, has several undesiderable practical consequences: (1) It may lead to under or overtreatment; (2) It has profound psychological and socioeconomic effects; (3) It greatly affects tumor incidence figures; (4) It may lead to a loss of confidence in the pathologist's competence on the part of the clinician; (5) It may lead to inconsistent and sometimes absurd terminology situations, in which tumors of similar morphology and behavior are called benign in one organ and malignant in another; (6) It causes an inordinate amount of anxiety for the pathologist. Elliot Foucar, an iconoclastic pathologist, has gone as far as suggesting that "the pathologists' inability to move beyond their benign vs. malignant paradigm should result in a loss of their terminology monopoly". [5] A thoughful reflection on the issue here discussed is contained in the book appropiately titled "The Histology of Borderline Cancer, with Notes on Prognosis", written by W. Wallace Park in 1980. [6] In the introductory chapter, entitled "A philosophy of cancer diagnosis by microscopy", he states: "Often enough, the pathologist is thought to dispense absolute truth: He does his best, but the truth he dispenses is not absolute. Rather, with a greater or lesser degree of confidence, he is making a prediction or expressing a probability of a certain kind of biologic behaviour. Every tissue pattern can be given a mathematical expression of risk: The odds are just more easily calculated in some than in others". Actually, the existence of a middle ground between those tumors of a totally indolent nature and those endowed with a highly aggressive behavior has been indirectly acknowledged over the years (sometimes tongue-in-cheek) by the use of terms such as semimalignant, grade 1/2 sarcoma, borderline, atypical, intermediate, and of uncertain malignant behavior. Terms expressing this philosophy first took hold in gynecologic pathology, (ovarian borderline serous tumors), but were eventually adopted for other organ systems, including soft tissue tumors. A good example of this development is that of adipose tissue tumors. They had been divided into benign (lipoma) and malignant (liposarcoma) in the classical scheme of Enzinger and Winslow, [1] but it was obvious to them that the latter category included lesions ranging from slow-growing, non metastasizing tumors to rapidly-growing, often metastasizing neoplasms. The ackward nature of this situation was made evident by the fact that what seemed basically to be the same tumor type was designated as well-differentiated liposarcoma by some authors and as atypical lipoma by others. In reality, both terms failed to express accurately the biologic potential of this distinctive tumor type, hence Harry Evans' proposal to forego the terms lipoma and liposarcoma for this particular tumor type and to replace it by the term "atypical lipomatous tumor", [2] a proposal later validated by the WHO Committee for the Classification of Tumors of Soft Tissue and Bone. [4] An analogous solution was proposed for a subtype of (uterine) smooth muscle tumors, which were referred to as smooth muscle tumors of uncertain malignant potential (STUMP), a subset of bladder tumors (papillary urothelial neoplasms of low malignant potential, or PUNLMP), and a subset of thyroid neoplasms (follicular tumors of uncertain malignant potential, or FT-UMP). I believe that the proposal of the existence of a borderline category among the various histotypes of soft tissue tumors represented an important conceptual advance. Unfortunately, its practical implementation left a lot to be desidered, the end result being that the term "borderline" was applied for tumors featuring a widely differing behavior, including the capacity for distant metastases. Perhaps for this reason, the WHO Committee has opted for the term "intermediate" in preference to "borderline", and has subdivided the tumors belonging into that category into "locally aggressive" (such as desmoid fibromatosis) and "rarely metastasizing" (such a plexiform fibrohistiocytic tumor and so-called angiomatoid fibrous histiocytoma). In any event, I see the introduction of the "borderline" concept (or whatever synonym is used for it) as the first step towards a more rational classification of soft tissue tumors. A much bigger step in the right direction was the proposal for the classification of gastrointestinal stromal tumors (GISTs) made by a Consensus Group sponsored by the NIH, [3]in which those tumors are not longer divided into benign or malignant (with or without a borderline category) but rather grouped according to their predicted risk for recurrence or metastasis into four categories: very low risk, low risk, intermediate risk, and high risk. One could argue about the criteria chosen (such as the lack of regard for mitotic activity), but not about the soundness of the concept, which delivers exactly what Wallace Park [6] had asked for almost 30 years ago, i.e., providing the clinical with an educated estimate of the tumor's expected behavior, without the use of the catchwords "benign" and "malignant". Significantly, the proposal gained immediate wide acceptance on the part of the oncologists, to the point that some of them now criticize the pathologist if they see the word "malignant" attached to a diagnosis of GIST (even if the tumor happens to be metastatic to the liver!). The obvious next question is: Now that we have finally made the big paradigmatic switch from benign to malignant (with the latter addition to borderline) to a risk assessment model for a particular tumor type, and we have found it good, what could stop us from exploring the some approach with other tumor types? Categories that seem ideally suited to this approach are solitary fibrous tumor/hemangiopericytoma, smooth muscle tumors (including those in the uterus), myxofibroblastic tumors, fibroblastic and myofibroblastic tumors, synovial sarcomas, phylloides tumors of the breast, and paragangliomas, What we need is for each of these tumor categories to be tackled by a expert committee like the one that dealt with GISTs, hopefully backed by a similar type of sponsoring. It goes without saying that this will be an arduous endeavor, which will need to take into account the peculiarities of each tumor category and probably incorporate data obtained from molecular biologic and other techniques. However, if successful, it is likely to become one of the greatest advances that will have been made in the understanding and handling of these neoplasms. References Enzinger FM, Winslow DJ. Liposarcoma. A study of 103 cases. Wirchows Arch Pathol Anat 335:367-388,1962. Evans HL. Liposarcoma. A study of 55 cases with a reassessment of its classification. Am J Surg Pathol 3:507-523,1979. Fletcher CDM, Berman JJ, Corless C, Gorstein F, Lasota J, Longley BJ, Miettinen M, O'Leary TJ, Remotti H, Rubin BP, Shmookler B, Sobin LH, Weiss SW. Int J Surg Pathol 10:81-90,2002; Hum Pathol 33:459-465, 2002. Fletcher CDM, Unni K, Mertens F. Tumors of Soft Tissue and Bone, Lyon, World Health Organization Classification of Tumours – Pathology and Genetics, 2002, pp. 35-37. Foucar E. Carcinoma-in-situ of the breast: have pathologists run amok? Lancet 16:707-708, 1996. Wallace Park W. The Histology of Borderline Cancer, With Notes on Prognosis. Berlin, Springer-Verlag, 1980, pp. 1-12.