Recently, my colleagues published two Nature papers. Nature is a weekly commercial scientific magazine recognized as one the most respected scientific journals and therefore publishing within it is recognized as a great scientific achievement. In their papers, they demonstrated that the geographical information of cancer cells, immune cells and other surrounding cells within a 1mm square of tumour core samples carries the information of patients’ prognosis. i.e. survival. The importance of tumour microenvironment (TME) has been emphasized for the last decade in cancer research. Now, with highly sophisticated cutting-edge technologies like image mass cytometry and artificial intelligence in their studies, its critical importance on cancer progression is further supported.
Cool. Then, their work ignited a very simple question in my mind. How can one 1mm square of one tissue section carry such information? A malignant tumour is a three-dimensional mass. To be detected by various non-invasive scanning technologies, it should be a visible size, not a microscopic size. This means that a tumour should be at least an approximately 1 cm cube in size. Realistically, 2-3 cm cubes in size would be often called as very early stages. Although those tumours are called early because of its size, the surgeons cannot conclude if the tumour is detected early enough for the complete resection leading to the patient’s cure. Recent studies suggest that the timing of metastasis, disseminating malignant cells to various parts of body, could be very early in some types of cancer. Even invisible in the current best scanning technologies, a small number of malignant cells might have spread and hidden in the patient’s body. If they already spread to unknown places around the body, it is impossible to resect all out by surgery. The possibility of recurrence is high if this happens. This means the prognosis of the patient is tough. How can just one 1 mm square of one section of a tumour core at an uncertain stage permit to predict the prognosis of a patient?
This does not make sense if you believe the current view of cancer progression. The dominant theory in the field of cancer research is the somatic mutation theory (SMT). Spontaneous advantageous mutations for proliferation and cell survival are gradually selected in a population. This permits to form a clonal population, a progeny derived from a single cell. Since DNA replication and cell division are the phases that have higher risks to introduce spontaneous mutations in the genome, the cells frequently entering cell-cycle have a higher probability to further accumulate mutations. Errors and selection enrich the cells carrying advantageous mutations for proliferation and survival.
How do they gain the traits for malignancy, such as invasion and metastasis? In the current view, the cells keep evolving into invasive metastatic cells by gaining more mutations. Invasion and metastasis are considered as the advantageous traits for proliferation and cell survival. The bad consequence of the constant evolution of malignant cells by mutations and selection. Consistent with this idea, the clonal heterogeneity is well recognized within many cancer types as intra-tumour heterogeneity. What this tells us is that the distribution pattern of clones within a tumour is varied and complex. This means that the clones within one 1 mm square of one section can be different from the clones within another 1 mm square of the same section.
The heterogeneity of gene expression in malignant cells and cancer-associate stromal cells is also well recognized. This information helps sub-classification of cancer types. Interestingly, neither the mutation landscape patterns nor gene expression patterns collected from a big chunk of a tumour does not have a strong prognosis-predicting power. On the other hand, the geographical information of malignant cells and immune cells within one 1 mm square does, without identification of mutations and characterization of gene expression. Does this make sense? Not for me. What is happening here?
Please do not misunderstand here. Identification of mutations and characterization of gene expression have predictive powers to select any specific drug effective or not. Particularly, for the recent molecular targeting drugs used as precision medicine. However, for predicting patient’s survival, somehow the cellular geographical information from one 1 mm square works better.
In my view, this suggests that the current view of cancer progression is insufficient, and thereby incorrectly capture what malignancy is. The central issue is if invasion and metastasis are traits gained through mutational selection in cancer cells. Does the ability of invasion and metastasis provide any advantage for proliferation and cell survival? In the teleological view, are the goal of cancer cells and the direction of selection gaining the ability of invasion and metastasis?
Perhaps, it would be important to consider an alternative view. Although malignancy appears to be co-selected with the traits for proliferation and cell survival, it may not be. Rather malignancy can be considered as the consequence of intracellular stress responses against unregulated and promoted cellular activities such as replication, transcription and translation. This leads to the compromised organelle integrity such as nuclear membrane integrity, mitochondria stress and ER stress. It is also possible to recognize that the traits for malignancy are not necessarily advantageous for proliferation and cell survival. In this view, malignant cancer cells are not the winner of selection but the half-collapsed cells by errors and selection screaming to get helps using their own intrinsic stress-response system. Then, the neighbour cells are forced to respond to this screaming, like responding to sterile injuries. However, there is no dead cells to be cleared. In addition, no apparent foreign antigen.
I see that malignancy is the condition created by an interaction of the screaming half-collapsed cells with their neighbour cells that attempt to recover back to a normal homeostatic state but fail, because the screaming cells are unable to be eliminated due to them being self-cells instead of non-self. Even after starting to scream, errors and selection continue based on proliferation and cell-survival in those cells. The clonal/mutation variation can keep increasing, thus the clonal heterogeneity increases. Accordingly, the heterogeneity of gene expression can increase.
Thus, malignancy is the consequence of an intimate interaction of malignant cells with their neighbours. This creates the minimum unit of malignancy. This unit can be captured within one 1 mm square, unlikely being evolving (therefore it has predictive power that is not affected by timing of clonal evolution) but being scaling (therefore random one 1 mm square is predictive). The minimum unit of malignancy is microscopic in size and consisting of highly heterogenous cell populations. The best way to capture the property of this unit is the geographical information of various cell types. Once this unit is established, it keeps replicating and expanding. No fundamental change occurs during disease progression.
What is the first step of malignancy? In my analogy, screaming. In scientific words, non-cell autonomous activities. I would like to mention that cellular stress response is not the only mechanism. It has been shown that some mutations have unique pathways to activate non-cell autonomous activities. They are not mutually exclusive as both can occur simultaneously. By screaming, neighbors are forced to be engaged and start building tumour microenvironment (TME). Does TME need to enhance the proliferation and survival of screaming cells? This is unnecessary. Even TME negatively impacts their proliferation and survival temporarily, they eventually adopt the new environment because of consistent errors and selection. The formation of TME is part of normal homeostatic responses for neighbours like responding to a minor sterile injury. But this minor sterile injury-like condition cannot be resolved because the screaming cells are unable to be eliminated. Thus, the condition continues like an unhealed wound.
Screaming should primarily be intrinsic to malignant cells. But the degree of screaming, loudness and frequency can be varied in each patient. In some cancer types, screaming can be softer and periodic while in others it can be intrinsically louder and continuous. In addition, TME can provide further external stresses. The external stresses enhance the level of screaming further louder and continuous, forcing more neighbors and even circulating immune populations to get involved. A local disease is progressing into a systemic disease. Probably, metastatic dormancy observed in some cancer types is because those malignant cells are intrinsically soft and periodic screamers without the external stresses. The level of external stresses is not great enough yet at the metastatic sites.
The minimum unit of malignancy is the intimate interaction of half-collapsed screaming cells with their neighbours at the microscopic scale. I think that the view of clonal selection leading to malignancy, i.e. SMT is insufficient. The view that ugly bad winner cells gaining evil superpowers to manipulate their environment as they like is not a good way to see this disease. Non-cell autonomous abilities are very difficult to be selected by errors and selection but are essential for malignancy. Recognizing that one 1 mm square from one section carrying the prognostic information is very important and provides new perspectives to challenge cancer.
I thank Warwick Pitman and Dr. Dardan Konjusha for English editing.