Lung cancer, like other types of cancer, is a very heterogeneous disease and not all patients respond in the same way to treatments such as immunotherapy; even within the same patient, some tumors do not react to immunotherapy, while others may even trigger unwanted autoimmune side effects (known as irAEs). For this reason, there is an urgent need to identify biomarkers and predictive models that anticipate which patients will benefit from certain treatments, thus facilitating personalized therapeutic strategies.

Age and the process of cellular aging (senescence) also play an important role in cancer, and this has strong scientific support; on the one hand, cellular senescence refers to a state in which cells permanently stop dividing but remain metabolically active, often releasing inflammatory signals and other factors into the environment. In addition, various mechanisms of aging, such as oxidative stress, cellular dysfunctions, or telomere shortening, can contribute both to tissue deterioration and to the appearance of cancer.

Previous studies have shown links between cellular senescence and tumor progression: for example, certain “aged” (senescent) immune cells can facilitate invasion and metastasis in laboratory models. Despite this, until now there were few studies that integrated multiple senescence-related genes to understand their combined influence in lung cancer, until in 2024 a study developed in hospitals in China explored how some aging-associated genes can be used to predict the prognosis of lung cancer and patients’ response to immunotherapy. The researchers integrated multi-omics analyses (that is, large-scale biological data from different sources) together with machine learning tools to build a prognostic model based on aging-related genes.

In essence, the goal was to identify senescence-related genes relevant in lung cancer and combine them into a “risk” index that could distinguish patients with a higher or lower probability of having an unfavorable clinical outcome and of responding well to immunotherapy. As a result, this analysis identified four key genes: BLK, ITGB4, PRKCH, and SNAI1 (genes involved respectively in B lymphocyte functions, cell adhesion, intracellular signaling, and cancer cell migration), with which they generated a “risk score”: a formula that sums the contribution of each gene to estimate the individual risk of each patient. In this way, each individual was assigned a score according to their genetic profile; higher values of this score were associated with less favorable prognoses (lower observed survival). The model was validated by dividing the data into training and testing cohorts, and even tested in an independent set of patients; in all cases it was confirmed that the risk score consistently separated patients into high- and low-survival groups. In addition, this risk index proved to be an independent prognostic factor, that is, it could predict the patient’s outcome even when considering other traditional clinical variables such as tumor stage or size.

Figure 1. Normal cells turn senescent, four aging-related genes shape tumor behavior, and AI uses this information to classify tumors as “cold” or “hot,” predicting their response to immunotherapy.

To better understand why the signature of these four genes is associated with differences in prognosis, the authors examined the biological functions and cellular pathways linked to the model. Through functional enrichment analysis, they observed that the model genes participate in immunological processes (for example, antigen binding and immune cell receptor signaling) as well as in functions related to cell proliferation and structure. In particular, they highlighted pathways such as the cell cycle, focal adhesion (cell-matrix interaction), cellular senescence, and p53 signaling (a crucial tumor suppressor pathway) among the most enriched when comparing high-risk versus low-risk tumors.

A key finding of this study is that the “risk score” defined by certain genes allows distinguishing how active a patient’s immune response is against their tumor. Patients with a high score, considered high-risk, showed a much weaker immune defense. Their tumors had less ability to destroy cancer cells and, in addition, presented a lower presence of lymphocytes and other immune cells. In these cases, fewer B cells, different types of T cells, specialized dendritic cells, and mast cells were found. This contrasts with low-risk tumors, which were surrounded by a more numerous “army” of immune defenses. In simple terms, high-risk tumors appear “cold” or invisible to the immune system, while low-risk tumors are more “hot” and more closely monitored.

Consistently, when reviewing key molecules that are usually the target of immunotherapy drugs, it was seen that low-risk tumors had higher levels of PD-1 and CTLA-4, two proteins targeted by modern drugs that unlock the immune system to act against cancer. In addition, when a global measure of immunotherapy response was evaluated, low-risk patients obtained better results. This suggests that, in practice, these patients could benefit more from immunological treatments than those with a high-risk score.

To explore this phenomenon further, the researchers turned to a cutting-edge technique called single-cell sequencing, which allows analyzing gene activity cell by cell within the tumor. With this approach, they observed that some genes were activated differently depending on the cell type. For example, ITGB4 was more present in tumor cells and blood vessel cells, while PRKCH and SNAI1 were mainly found in endothelial cells. In contrast, BLK, a gene characteristic of B lymphocytes, was expressed at very low levels in high-risk tumors, reflecting the scarcity of these cells in those patients. This finding fits perfectly with the previous observation that high-risk tumors had less B cell infiltration.

Taken together, these four genes (BLK, ITGB4, PRKCH, and SNAI1) form a “genetic signature” that reflects two things at the same time: on the one hand, how the tumor behaves in terms of growth, adhesion, and invasion capacity; and on the other, how present and active the immune system is around the tumor. Thanks to this model, which combines genomic data and artificial intelligence tools, in the future doctors could have a method to predict more precisely which patients have more aggressive tumors and who could respond better to immunotherapies. This type of advance opens the door to much more personalized medicine in the treatment of lung cancer, adapting therapies to the unique biological characteristics of each patient.

Main Reference:

Lu, W., Zhou, Y., Zhao, R., Liu, Q., Yang, W., & Zhu, T. (2024). The integration of multi-omics analysis and machine learning for the identification of prognostic assessment and immunotherapy efficacy through aging-associated genes in lung cancer. Aging, 16(2), 1860–1878. https://doi.org/10.18632/aging.205464

Other References:

Nasim, F., Sabath, B. F., & Eapen, G. A. (2019). Lung Cancer. The Medical clinics of North America, 103(3), 463–473. https://doi.org/10.1016/j.mcna.2018.12.006

Bade, B. C., & Dela Cruz, C. S. (2020). Lung Cancer 2020: Epidemiology, Etiology, and Prevention. Clinics in chest medicine, 41(1), 1–24. https://doi.org/10.1016/j.ccm.2019.10.001

Chaft, J. E., Rimner, A., Weder, W., Azzoli, C. G., Kris, M. G., & Cascone, T. (2021). Evolution of systemic therapy for stages I-III non-metastatic non-small-cell lung cancer. Nature reviews. Clinical oncology, 18(9), 547–557. https://doi.org/10.1038/s41571-021-00501-4