The Biology Of Cancer

Understanding How Tumors Actually Grow

I spent four years running cell lines in a lab that smelled like formaldehyde and old coffee. Most of what I learned about cancer didn't come from textbooks. It came from watching cultures get contaminated at 2 AM and trying to figure out why a drug that worked in vitro completely failed in vivo. The Biology Of Cancer is fundamentally about cells that stopped listening to their neighbors. That's the core of it. Everything else is details about how that happens. Cancer isn't one disease. It's a set of failure modes in multicellular cooperation. Normal cells in your body follow rules: divide when told to, stop when crowded, repair DNA damage or self-destruct, migrate only to specific locations. Cancer cells break these rules through accumulated mutations. The rate at which this happens varies wildly between tissue types. Skin gets more UV damage. Lungs get more carcinogens. Colon cells divide frequently, giving more opportunities for copying errors.

The Biology Of Cancer and Why Some Treatments Fail

Here's something most introductory material gets wrong. Cancer doesn't start because of one bad mutation. It starts because of a combination of mutations that together disable tumor suppression and enable uncontrolled growth. The classic model involves approximately six to eight key driver mutations across genes like TP53, RAS, MYC, and PTEN. But the exact combination varies from patient to patient. Two people with lung cancer will almost never have the identical mutational profile. The counterintuitive part is that more mutations don't always mean worse cancer. Some tumors with massive mutational loads respond better to immunotherapy because they present more neoantigens for the immune system to recognize. A tumor with fewer mutations can actually be more dangerous because it's harder for your immune system to spot. This is why melanoma and lung cancer patients often respond to checkpoint inhibitors while pancreatic cancer patients largely don't. Pancreatic tumors have fewer mutations and a denser stromal barrier that blocks immune cell access. I ran into a specific problem during a project involving targeted therapy resistance. We were tracking how breast cancer cells with HER2 amplification responded to trastuzumab. After about eight weeks of treatment, the sensitive cell line developed resistance. Standard protocol would suggest doing another round of sequencing to find a new mutation. Instead, I noticed something odd under the microscope. The resistant cells looked different. They were larger, flatter, and had lost their typical adherent growth pattern. They had undergone epithelial-to-mesenchymal transition, basically changing their cell identity rather than evolving a new mutation. The workaround was combining the HER2 inhibitor with a pathway blocker targeting the TGF-beta signaling that was driving the EMT. That combination resensitized the cells within two weeks. Sequencing alone would have missed this entirely because the genome hadn't changed. Only the epigenome and the phenotype had.

How Metastasis Actually Works

Metastasis is the reason most cancer deaths happen. The primary tumor itself is often removable. It's the cells that break away, travel through blood or lymph, and establish secondary tumors that are lethal. The process is brutally inefficient. Of the cells that enter the circulation, fewer than one in a million survives to form a detectable metastasis. The immune system destroys most of them. The physical stress of circulation damages them. Many just can't survive in a foreign tissue environment. For a metastasis to establish, the circulating tumor cell needs to colonize a new organ. This isn't random. There's a reason breast cancer commonly spreads to bone, lung, liver, and brain, and rarely to muscle. The "seed and soil" hypothesis from 1889 was basically correct, though we now understand it in molecular detail. Organ-specific microenvironments express different sets of chemokines and growth factors. Tumor cells evolve preferences for certain tissues. Breast cancer cells that express the chemokine receptor CXCR4 are drawn to bone marrow, which secretes the corresponding ligand CXCL12. This homing mechanism is specific enough that we can map likely metastatic sites by looking at receptor expression patterns in the primary tumor. One thing people don't appreciate is how long metastasis can remain dormant. Cells can circulate and lodge in distant organs and stay there for years without dividing. I've seen cases where patients were considered cured after five years and then developed metastatic disease at year twelve or fifteen. The dormant cells are usually in a quiescent state, kept in check by the local microenvironment. Stress, inflammation, or changes in the niche can wake them up. We don't have good ways to predict which dormant cells will reactivate. This is one of the biggest unsolved problems in clinical oncology.

What Happens Inside a Tumor Microenvironment

Tumors aren't just masses of cancer cells. They're complex organs made of cancer cells, immune cells, fibroblasts, blood vessels, and signaling molecules all interacting in messy ways. The tumor microenvironment can suppress immune responses, promote angiogenesis, and even help cancer cells resist chemotherapy. About 30 to 40 percent of a tumor's mass can be non-cancerous cells and extracellular matrix. Cancer-associated fibroblasts are particularly important. Normal fibroblasts maintain tissue structure. Cancer-associated fibroblasts, or CAFs, get reprogrammed by signals from the tumor cells to produce dense collagen deposits, secrete growth factors, and release immunosuppressive cytokines. They essentially build a wall around the tumor that excludes immune cells and makes drug delivery harder. I worked on a project where we tried to improve chemotherapy penetration by using an enzyme called PEGPH20 to degrade hyaluronic acid in the tumor stroma. It worked in early trials for pancreatic cancer, reducing interstitial pressure and allowing more drug into the tumor. But the benefit only lasted a few weeks before the stroma regenerated. And the trial was eventually halted because of thrombotic complications. The stroma isn't just a barrier. It's a dynamic system that responds to being disrupted. The immune component of the microenvironment is equally complex. Tumors recruit regulatory T cells, myeloid-derived suppressor cells, and M2 macrophages that actively suppress anti-tumor immunity. Checkpoint inhibitor therapies work by releasing the brakes on T cells, but they only help patients whose tumors already have some immune infiltration. We call these "hot" tumors. "Cold" tumors with no immune cells generally don't respond to checkpoint inhibitors regardless of mutation load. Getting immune cells into cold tumors is an active area of research involving oncolytic viruses, STING agonists, and radiation-induced antigen release.

Practical Approaches to Studying Cancer Biology

If you're working with cancer cell lines, the first thing to know is that most established lines are badly misidentified or contaminated. A 2012 analysis found that something like 18 to 20 percent of cell lines in use were not what they claimed to be. HeLa contamination alone has infected thousands of cultures worldwide. Always validate your cell lines with STR profiling before drawing conclusions from experiments. It takes about two days and a few hundred dollars. Skipping it has invalidated entire published studies. Three-dimensional culture models are significantly better at predicting drug response than traditional two-dimensional monolayers. Spheroids and organoids maintain cell-cell interactions and gradients of oxygen and nutrients that exist in real tumors. Drugs that look effective in 2D culture often fail in 3D because they can't penetrate the spheroid core or because the hypoxic center induces a completely different gene expression program. Growing spheroids adds maybe one day to your workflow but improves the physiological relevance dramatically. Organoids are even better but require specialized media, take longer to establish, and are more expensive. When moving to animal models, note that immunocompromised mice used for xenografts don't have a functioning immune system, so drug effects you observe may be entirely different from what happens in a patient. Patient-derived xenografts are more predictive but take months to establish and only engraft about 50 to 60 percent of implantation attempts. For mechanistic studies, genetically engineered mouse models are superior but require significant expertise to generate and maintain. There is no single model that faithfully reproduces human cancer. Each has specific blind spots. The best approach is to use multiple complementary models and triangulate your findings.

Limitations and Where Current Understanding Falls Short

Despite decades of research, we still cannot reliably predict which precancerous lesions will progress to invasive cancer. A Barrett's esophagus patient, a patient with a dysplastic nevus, and a patient with ductal carcinoma in situ of the breast all have abnormal cells. We have no reliable way to tell which ones will become lethal. This is a fundamental gap in the field. We can detect cancer after the fact with increasing sensitivity. We can treat it once it's there with variable success. But we cannot accurately forecast progression from pre-malignant states in individual patients. Another limitation is our understanding of cancer stem cells. The idea that a small subpopulation of cells within a tumor drives recurrence and metastasis has some supporting evidence but remains controversial. Single-cell RNA sequencing shows heterogeneity within tumors, but whether that heterogeneity is driven by a hierarchy of stem-like cells or is mostly plastic and context-dependent is still debated. The clinical implication is that even if you kill 99.9 percent of a tumor, the remaining cells might regenerate the entire mass if they include the right subpopulation. This is why adjuvant therapy is standard after surgical resection, even when margins are clear. The biggest practical limitation is that most preclinical findings don't translate to clinical success. Drug development pipelines in oncology have notoriously low hit rates. A compound that shrinks a tumor in a mouse model has roughly a 5 percent chance of reaching approval in humans. The biological complexity we've been discussing is exactly why the translation gap exists. Tumors evolve during treatment. Microenvironments differ between species. Dosing schedules optimized for mice don't necessarily match what humans can tolerate. There's no shortcut around this except rigorous preclinical work across multiple models and species before moving to patients.