This is one of the most insightful books on cancer that I've read, and books like this are sorely needed, given our lack of progress in dealing with cancer over the past few decades (despite vast expense and endless media hype). Every cancer researcher and clinical oncologist needs to read it without delay.
Berman's background is unusual in its multidisciplinary breadth and depth, and this puts him in a unique position to comprehensively examine and synthesize our current biological and clinical knowledge regarding cancer. The resulting 400-page book is rich with important details, yet not difficult to read for people with biomedical background. To give a sense of Berman's findings, the following are some of the key points, but be sure to read the entire book, since no summary can do it justice:
(1) Cancer is found in virtually all multicellular organisms, including both plants and animals, although it's less common in insects because most cells in adult insects are post-mitotic.
(2) Tumor cells are abnormal in countless ways (especially with regard to their nuclei), every tumor is somewhat unique, every tumor cell is somewhat unique, and tumors are generally much more complex than the normal tissues from which they derive. And yet, despite this heterogeneity, there are about 6,000 distinct types or "species" of tumors - a large but finite number - and newly reported types of tumors don't arise in human populations. Moreover, only 17 tumor types account for over 90% of human tumors, thus making most tumor types relatively rare.
(3) Based primarily on developmental lineage (determined by epigenetic regulation of expression of pre-existing normal pathways to determine phenotype) and secondarily genetic and molecular features, these thousands of tumor types can be grouped into six classes (endoderm/ectoderm, mesoderm, neuroectoderm, neural crest, germ cell, and trophectoderm), and each of these six classes can be further divided into subclasses, resulting in about 40 total subclasses. The resulting tumor classification is hierarchical, exhaustive (covers all tumor types, including every stage of tumor development), unique (each tumor falls into only one subclass and can't transition to another subclass), can readily be related to the currently used tumor taxonomy, and hopefully is useful for treatment purposes (a given treatment may be effective for all tumors within a subclass).
(4) The class of endodermal/ectodermal tumors accounts for more than 95% of all human cancer incidence, with just six of these tumor types accounting for about 60% of all human cancer mortality, generally later in life. This is mainly because endodermal/ectodermal tissues have greater exposure to carcinogens due to surface exposure, they are the greatest metabolic activators of carcinogens to their mutagenic forms, and they have the greatest rates of mitosis. These tumors rarely have just a few simple gene alterations, and instead tend to be aneuploid, heterogeneous, and (epi)genetically unstable, which makes them especially difficult to treat effectively.
(5) All malignant tumors grow, persist, invade, and metastasize, and the malignant phenotype typically develops in this sequence. Normal tissues do each of these also, but no normal tissue does all four together. Carcinogenesis involves both genetic and epigenetic alterations, typically involving oncogenes and tumor suppressor genes, with development of tumor suppressor genes usually preceding oncogenes. Usually only one oncogene is active in a tumor cell and particular oncogenes are specific to particular tumor types. By contrast, multiple tumor suppressor genes may be active in a tumor cell, and they're generally not as specific to particular tumor types; tumor suppressor genes are involved in cell cycle arrest/control, apoptosis, DNA repair, and (epi)genomic instability, and they alone don't generally produce a malignant phenotype.
(6) The carcinogenic process is likely monoclonal, appears to typically involve multiple events (specific causes usually can't be determined in particular patient cases), and can take weeks (very rare) to years (the norm), with the steps being initiation, latency, precancer (non-invasive), malignant cancer (invasive and possibly metastatic), and progression. Except in cases of spontaneous regression (which is most common with precancer), progression generally continues and typically involves some combination of increasing aneuploidy, increasing tumor heterogeneity, metastasis, dormancy, dedifferentiation, and accelerated tumor growth. Progression is due to (epi)genetic instability, and results in some clones outcompeting others. Progression can also arise suddenly in indolent and benign tumors which were previously impaired by low (epi)genetic instability (and thus lacked growth and invasiveness).
(7) The details of carcinogenesis are still largely a mystery, and thus there are various competing theories, such as the dominant somatic mutation theory, along with field theory and stem cell theory. The field theory (eg, tissue organization field theory) posits that carcinogens poison the stroma, which causes disordered tissue growth, followed by (epi)genetic changes and tumor growth. The tumor stem cell theory posits that tumors derive from stem cells which renew themselves while also producing differentiated cells which form the bulk of the tumor. There's evidence for and against all of these theories, and there may actually be multiple different carcinogenic processes which can act in varying combinations and differ from one tumor type to another, so it may not be possible to develop a single simple theory which covers all cases. The situation is similar with the theory of immunosurvelliance of cancer.
(8) Tumors can also be classified based on their complexity, and most tumors are either relatively simple or relatively complex (ie, towards the extremes). Simpler tumors tend to be relatively rare, observed during childhood (only 0.4% of all cancer mortality), caused by a single biological alteration (and thus relatively homogeneous), more (epi)genetically stable, less likely to be metastatic, and more readily curable (eg, by targeting a single pathway). By contrast, complex tumors tend to be much more common, observed during later adulthood, caused by multiple biological alterations (and thus much more heterogeneous), more (epi)genetically unstable, advanced (invasive and possibly metastatic), and rarely curable once they've reached advanced stage. Also, simple and complex tumors tend to arise from different tissues.
Perhaps the most important clinical implication is that understanding the developmental biology of tumors could give insights regarding which pathways to target in complex tumors. However, it remains to be seen whether a multi-targeted approach along these lines can overcome the redundancy, adaptability, and evolvability of complex tumors without excessive toxicity. Personally, I'm not highly optimistic, but of course I hope to be proven wrong.