Lung Cancer Histology: Types, Markers, and Treatment

Lung cancer is not a single disease. Under the microscope, tumors fall into distinct histological types that differ in how they look, which genes drive them, and how they respond to treatment. The broadest split is between non-small cell lung cancer (NSCLC), which accounts for roughly 85% of cases, and small cell lung cancer (SCLC), which makes up most of the rest. Within NSCLC, further subdivision into adenocarcinoma, squamous cell carcinoma, and large cell carcinoma steers nearly every decision about molecular testing, drug selection, and prognosis. Getting the histology right is not academic bookkeeping; it is the foundation on which modern lung cancer treatment is built.

Adenocarcinoma and Why Its Subtypes Matter

Adenocarcinoma is the most common form of lung cancer overall and the dominant type in people who have never smoked. It typically arises in the outer regions of the lung from glandular cells that produce mucus. What makes adenocarcinoma especially interesting from a histology standpoint is that it is not one uniform pattern. The tumor can grow in several architectural styles, and which pattern predominates tells doctors a great deal about likely outcomes.

A large validation study of the international adenocarcinoma classification found that median overall survival varied substantially by predominant growth pattern: about 78.5 months for lepidic-predominant tumors, 67.3 months for acinar, 58.1 months for solid, 48.9 months for papillary, and 44.9 months for micropapillary-predominant tumors. These differences held regardless of stage or treatment received.¹ A separate study confirmed that a solid growth pattern in particular signals worse outcomes, acting as a marker of unfavorable prognosis in primary lung adenocarcinoma.²

In practice, pathologists examine stained tissue slides and classify the predominant pattern. A tumor might have several patterns mixed together, but the one that dominates drives the prognostic grouping. Lepidic-predominant tumors tend to grow slowly along existing lung structures and carry the best prognosis, while micropapillary tumors are notorious for spreading aggressively. This is one reason a tissue biopsy large enough for a pathologist to assess architecture is so valuable, and why a tiny needle aspirate sometimes does not tell the full story.

Squamous Cell Carcinoma

Squamous cell carcinoma (SCC) of the lung historically arose in the central airways and was tightly linked to heavy smoking. It develops from the flat cells lining the bronchi and tends to be a genetically complex tumor. A landmark genomic study profiling 178 lung squamous cell carcinomas found a mean of 360 exonic mutations, 165 genomic rearrangements, and 323 segments of copy number alteration per tumor.³ That level of genomic chaos has made it harder to find single targetable driver mutations in squamous tumors compared to adenocarcinomas.

This distinction matters for treatment. While adenocarcinomas frequently harbor “druggable” mutations like EGFR or ALK rearrangements, squamous tumors rarely do. Some drugs used in adenocarcinoma, such as certain anti-angiogenic agents, carry unacceptable bleeding risks in squamous histology due to the tendency of these tumors to involve large central blood vessels. Knowing whether a tumor is squamous or adenocarcinoma therefore directly influences which chemotherapy backbone, targeted therapy, or immunotherapy regimen a patient receives.

Small Cell Lung Cancer

SCLC is a different beast altogether. It is an aggressive, poorly differentiated, high-grade neuroendocrine carcinoma that accounts for about 13% of all lung cancers and is the most common neuroendocrine lung tumor.⁴ Almost all cases occur in heavy smokers. SCLC cells are small with scant cytoplasm, and they divide rapidly. The disease is usually widespread at diagnosis, which is why it is staged simply as “limited” or “extensive” rather than using the detailed TNM staging applied to NSCLC.

SCLC initially responds well to chemotherapy and radiation, sometimes dramatically so, but it almost always relapses within months, and second-line options have historically been limited. The biology behind this aggressiveness involves near-universal loss of the tumor suppressor genes TP53 and RB1, which removes the two main brakes on uncontrolled cell division.

The Neuroendocrine Spectrum

SCLC sits at the high-grade end of a broader family of neuroendocrine lung tumors. The spectrum ranges from low-grade typical carcinoid and intermediate-grade atypical carcinoid to high-grade large cell neuroendocrine carcinoma (LCNEC) and SCLC. Carcinoids and high-grade tumors are fundamentally different diseases despite sharing neuroendocrine features: carcinoids are not strongly linked to smoking and carry relatively few genetic abnormalities, while SCLC and LCNEC occur in older heavy smokers and harbor extensive genetic damage.⁵

LCNEC is a particularly tricky diagnosis. It has large cells with neuroendocrine features but behaves aggressively like SCLC. Studies looking at how well pathologists agree on the diagnosis of pulmonary carcinoids and LCNEC have found only modest concordance, with unanimous diagnoses at a median of about 59% and interobserver agreement scores ranging widely.⁶ This means expert review and molecular testing can be especially important for these rare subtypes.

How Pathologists Tell Them Apart

When tissue is obtained through biopsy, pathologists first look at cell shape and growth patterns on standard stained slides. But in small biopsies or crushed specimens, architecture can be ambiguous. That is where immunohistochemistry (IHC) comes in. IHC uses antibodies to detect specific proteins in the tissue, and a streamlined panel of just two markers can resolve most NSCLC cases.

TTF-1 (thyroid transcription factor 1) lights up in most adenocarcinomas, while p40 lights up in squamous cell carcinomas. A study evaluating this two-marker panel found that TTF-1 identified adenocarcinoma with a sensitivity over 97% and a positive predictive value of 100%, while p40 detected squamous cell carcinoma with both sensitivity and positive predictive value at 100%.⁷ A separate cross-sectional study confirmed that TTF-1 has high sensitivity and specificity for primary lung adenocarcinomas and that p40 outperforms the older marker p63 for identifying squamous differentiation.⁸

Keeping the IHC panel minimal is not just about convenience. Biopsy tissue is precious, and every stain uses up material. A lean panel of TTF-1 and p40 conserves tissue for molecular testing of targetable driver mutations, which increasingly determines treatment. For neuroendocrine tumors, pathologists add markers like chromogranin, synaptophysin, and CD56, which confirm the neuroendocrine character of SCLC and LCNEC.

An unusual wrinkle involves tumors that co-express both TTF-1 and p40, markers that normally mark opposite subtypes. A recent study found that these dual-positive tumors represent a distinct group with frequent TP53 mutations and dysregulation of the FGFR signaling pathway.⁹ These cases challenge the neat adenocarcinoma-versus-squamous binary and may eventually require their own treatment approach.

Oncogenic Driver Mutations and Targeted Therapy

One of the biggest advances in lung cancer treatment has been the discovery that many adenocarcinomas are driven by specific, identifiable gene alterations, each of which can be attacked with a matched drug. The most common of these are EGFR mutations, ALK rearrangements, and ROS1 rearrangements. In a large series from South India, EGFR mutations were found in about 34% of NSCLC patients, ALK rearrangements in roughly 11%, and ROS1 rearrangements in about 2%. Among EGFR-mutated tumors, the two most frequent types were exon 19 deletions and the L858R point mutation, and both were overwhelmingly found in adenocarcinoma histology.¹⁰ Rates vary by geography and ethnicity, with EGFR mutations being more common in East Asian populations and less common in Western cohorts.

ALK rearrangements define a molecular subtype of NSCLC that responds remarkably well to ALK inhibitors. Crizotinib was the first approved agent for ALK-positive disease, and it was also recognized to have activity against ROS1 rearrangements.¹¹ Newer-generation ALK inhibitors like alectinib and lorlatinib have since displaced crizotinib as preferred first-line options due to better efficacy and brain penetration.

Beyond these well-established targets, genome studies have identified additional therapeutic targets in lung adenocarcinoma, including RET fusions, NTRK fusions, MET alterations, and activating mutations in KRAS, BRAF, and HER2, all occurring at frequencies above 1%.¹² Drugs for several of these are now approved or in advanced trials. KRAS G12C, for instance, was considered “undruggable” for decades before sotorasib and adagrasib reached the clinic. This expanding menu of targeted therapies is one reason comprehensive molecular profiling at diagnosis is now standard for advanced NSCLC.

How Smoking Shapes the Mutational Landscape

Lung cancers in smokers and never-smokers are genetically different diseases.¹³ A comprehensive genomic analysis found that tobacco smokers had a median of about 10.5 mutations per megabase of DNA, compared to a median of 0.6 in never-smokers, a difference of more than tenfold. The types of mutations also differed: smokers’ tumors were dominated by a specific kind of DNA damage signature (C:G→A:T transversions), while never-smokers showed a different pattern (C:G→T:A transitions).¹⁴

This has practical implications. Never-smoker adenocarcinomas are enriched for EGFR mutations and ALK rearrangements, making them more likely to benefit from targeted therapies. Smoker-associated tumors tend to have higher mutational burdens and more chaotic genomes, which can make them better candidates for immunotherapy, since more mutations can produce more abnormal proteins for the immune system to recognize.

PD-L1 and Tumor Mutational Burden as Immunotherapy Biomarkers

Immune checkpoint inhibitors have transformed treatment for many lung cancer patients, but not everyone benefits equally. Two biomarkers help predict who is most likely to respond. PD-L1 is a protein that tumors can display on their surface to suppress immune attack. Pathologists score PD-L1 expression using IHC, typically reporting the percentage of tumor cells that stain positive, known as the Tumor Proportion Score (TPS). A TPS of 50% or higher qualifies patients for first-line single-agent immunotherapy in many guidelines, while a TPS between 1% and 49% often leads to immunotherapy combined with chemotherapy.¹⁵

Tumor mutational burden (TMB) is the second major biomarker. It estimates how many mutations exist across the tumor genome and serves as a rough indicator of how many abnormal proteins the tumor produces, and therefore how visible it might be to the immune system. Higher TMB tends to correlate with better outcomes after immune checkpoint therapy across tumor types. TMB and PD-L1 are biologically related but function as independent predictors of immunotherapy response, meaning some patients with low PD-L1 still benefit if their TMB is high, and vice versa.¹⁶ A systematic review confirmed that most published studies found an association between high TMB and better progression-free survival and response rates with immunotherapy.¹⁷

PD-L1 scoring, however, has its frustrations. Pathologists can disagree on exact percentages, especially near the clinically important 1% and 50% thresholds. Automated scoring tools using digital pathology have shown strong correlation with manual pathologist assessments, with one study reporting concordance between automated and average pathologist scores at 90% for the 1% threshold and 92% for the 50% threshold.¹⁸ These tools may help standardize scoring over time.

Histological Transformation as a Resistance Mechanism

One of the more alarming phenomena in lung cancer biology is histological transformation, where a tumor changes its fundamental cell type under the selective pressure of treatment. The best-studied example involves EGFR-mutated adenocarcinomas that transform into small cell lung cancer after treatment with EGFR-targeted drugs like osimertinib. This transformation dramatically worsens prognosis because the new small cell phenotype is highly resistant to continued EGFR-targeted therapy.¹⁹

The transformed tumors retain the original EGFR mutation but have essentially switched their entire gene expression program, usually acquiring RB1 and TP53 loss along the way. Research into these transformed tumors has shown significant intratumoral heterogeneity, meaning that not all cells transform simultaneously. Some cells may remain adenocarcinoma-like while others have already become small cell, which complicates treatment decisions.²⁰ This is why re-biopsy at the time of treatment resistance is so important. Without a fresh tissue sample, clinicians might continue treating for adenocarcinoma when the tumor has already become something else entirely.

Liquid Biopsy and When Tissue Is Not Available

Traditional histological diagnosis requires tissue, usually obtained through bronchoscopy, CT-guided needle biopsy, or surgical sampling. But tissue biopsies are invasive, sometimes risky, and occasionally impossible if the tumor is in an inaccessible location. Liquid biopsy has emerged as a complementary tool: a blood draw that captures circulating tumor DNA (ctDNA) shed by the cancer into the bloodstream.²¹ This ctDNA can be analyzed for the same driver mutations detected on tissue, making it especially useful for treatment selection and monitoring resistance.²²

Liquid biopsy works best in advanced disease, where tumor DNA is more abundant in the blood. In early-stage disease, detection is much less reliable. A study testing for KRAS mutations in plasma samples found that in early-stage lung adenocarcinoma, mutations were detected in only one of fifteen samples processed from plasma, while in late-stage disease the detection rate jumped to thirteen out of fourteen.²³ Liquid biopsy cannot replace tissue for initial histological classification, since you need to see cell architecture under a microscope to determine whether a tumor is adenocarcinoma, squamous, or small cell. But it can pick up mutations, track treatment response over time, and detect emerging resistance without subjecting a patient to another procedure.

Antibody-Drug Conjugates

A newer class of therapy bridges the gap between targeted drugs and chemotherapy. Antibody-drug conjugates, or ADCs, consist of an antibody aimed at a protein on the tumor surface, connected through a chemical linker to a potent cytotoxic payload. When the antibody locks onto its target on a cancer cell, the whole complex gets pulled inside, and the toxic payload is released to kill the cell from within. Several targets in lung cancer are under active development, including TROP2, HER2, HER3, MET, and CEACAM5.²⁴

TROP2 is overexpressed in many lung cancers while remaining mostly absent on normal tissue, making it an attractive target. A TROP2-directed ADC has already been approved by the FDA.²⁵ Preclinical work has also shown that HER2- and TROP2-directed ADCs have activity in small cell lung cancer models, and that expression levels of these targets may increase after first-line chemotherapy, suggesting these drugs could be especially useful in relapsed disease where options are scarce.²⁶ ADCs are particularly exciting because they can deliver chemotherapy precisely to tumor cells, potentially reducing the collateral damage that conventional chemo inflicts on healthy tissue.

Artificial Intelligence in Lung Cancer Pathology

Classifying lung cancer histology under a microscope is skilled work, and some diagnoses are harder than others. Deep learning algorithms trained on digitized whole-slide images are now capable of distinguishing adenocarcinoma from squamous cell carcinoma from normal lung tissue with performance comparable to pathologists, achieving an average area under the curve of 0.97. The same computational approach was then used to predict the mutational status of commonly mutated genes directly from histology images, without any molecular testing, with six genes (including STK11, EGFR, and KRAS) predicted with AUCs ranging from 0.733 to 0.856.²⁷

A systematic review of the field found that most published deep learning studies focus on distinguishing adenocarcinoma, squamous cell carcinoma, and small cell carcinoma, but some also tackle finer-grained tasks like adenocarcinoma growth pattern classification, prognosis prediction, and even PD-L1 expression estimation from standard pathology slides.²⁸ These tools are not replacing pathologists, but they could function as quality-assurance aids, flagging cases where the algorithm and human disagree so that a second look can catch diagnostic errors.

Spatial Transcriptomics and the Tumor Microenvironment

Traditional histology tells you what a tumor looks like. Molecular testing tells you what mutations it carries. Spatial transcriptomics does something neither can do alone: it maps which genes are active in which cells, in their exact physical locations within the tissue. A recent study applied this technique to early-stage lung adenocarcinoma across histological subtypes and found that poorly differentiated tumors had distinct molecular activity in both their cancer cells and surrounding immune cells, with enrichment of pathways related to immune evasion and tissue remodeling that correlated with worse prognosis.²⁹

This kind of data could eventually refine prognosis and treatment beyond what histology or sequencing alone can offer. A tumor might look the same under a microscope as another tumor of the same subtype, yet have a very different immune microenvironment, meaning it would respond differently to immunotherapy. Spatial technologies are still largely research tools and expensive to deploy, but they represent the direction lung cancer diagnostics is headed: layering architecture, genomics, and microenvironmental context into a single integrated picture of each individual patient’s disease.