Isolation and identification of aneuploid CTCs and CTECs
Firstly, we removed the blood-derived cells such as red blood cells and white blood cells by centrifugation and phase-enrichment to achieve higher CTC enrichment efficiency, and then we borrowed the iFISH platform, which combine the immunofluorescence staining of tumor proteins (vimentin, KI67, PD-L1, etc.) with the detection of chromosomal aneuploidy. It can achieve the subclassification of CTCs with a higher sensitivity and specificity. The CTC identification criteria are as follows (Fig. 1A): nuclear DAPI+, CD45-, CD31-, chromosome 8 (CEP8) aneuploidy positive, and immunofluorescent staining CTC tumor markers positive or negative. Using this CTC sorting method, we successfully enriched and identified CTCs in peripheral blood of 98 patients (clinical characteristics are shown in Table 1). When screening aneuploid CTCs under a fluorescence microscope, we also found a large number of aneuploid endothelial cells (CTECs). These cells had the same degree of aneuploidy as tumor cells, but the endothelial cell marker CD31 was strongly positive (Fig. 1B)
Distribution of total CTC and CTEC numbers in patients by stage
We used the iFISH platform to enrich and identify CTCs and CTECs in 17 healthy volunteers, 22 benign pulmonary nodules, 34 early lung adenocarcinoma patients, and 22 advanced lung adenocarcinoma patients. We found that CTCs could be detected in the very early stages of lung cancer. Of the 34 early stage lung cancer patients, 29 cases detected the presence of CTCs, of which 17 cases had more than three CTCs. A certain number of CTCs could also be isolated in normal and benign nodule patients. This part of the results indicated that there was a certain background pollution in normal people and benign nodules (non-tumor cells of aging or apoptosis also had a certain aneuploidy, which was mistaken for tumor cells), but the number was mostly less than three, which indicated that CTCs still had certain advantages in identifying benign nodules and early stage lung cancer(P = 0.0119) (Fig. 2A). The number of CTCs in patients with advanced lung cancer was significantly increased. At the same time, the number of peripheral blood CTCs in patients with advanced lung adenocarcinoma was significantly higher than that in early lung cancer, benign nodules and normal controls, the difference was statistically significant (P < 0.0001) (Fig. 2A). In addition, we found some CTC-like aneuploidy circulating tumor endothelial cells (CTECs) during the experiment. Their quantitative changes in normal human, benign nodules and early and advanced lung cancer were consistent with CTCs (Fig. 2B). The number of CTECs in advanced patients was significantly higher than that of normal, benign nodules, and early stage lung cancer patients, and it had the same early diagnostic value (P = 0.0287) and tumor staging diagnostic value(P = 0.0009) as CTCs. Since there are still normal diploid endothelial cells (CECs) in peripheral blood, this part of the cells is reported to have a certain staged diagnostic significance[2–6], and closely related to survival[7] and relapse[8] of cancer patient. We also counted the number of diploid CECs and found the number of diploid CECs in peripheral blood was very few, therefore we combined diploid CECs and aneuploid CTECs to observe their dynamic changes. The results showed that the total CEC score is not as good as the aneuploid CTECs in identifying early and late lung adenocarcinoma(Fig. 2C).Unexpectedly, we found that there was a certain correlation between CTCs and aneuploid CTECs in the peripheral blood, and there was a statistically significant difference(Fig. 2E) (P = 0.0003). CTCs and aneuploid CTECs are divided into two categories only because of the positive or negative CD31. As two independent and interrelated tumor markers, we suspected that the combined diagnosis of the two might have a better effect. The results showed that CTCs and aneuploid CTECs combined with a single marker had a greater advantage in distinguishing between benign and malignant nodules, statistical differences were more obvious (Fig. 2D) (P = 0.0069).
Aneuploidy analysis of CTCs and CTECs
Our study found that both CTCs and CTECs exhibited varying degrees of chromosome aneuploidy, namely, haploid, triploid, tetraploid, pentaploid and above, standard images are shown in Fig. 3A. The clinical relevance of different types of aneuploid CTCs has been demonstrated, for example, high-ploid CTCs has been shown to be associated with tumor resistance and relapse[9, 10]. However, the clinical significance of aneuploid CTECs are still unclear.
We found that these sorted CTC chromosomes 8 had a certain pattern. In all normal human, benign nodules, patients with early-stage or late-stage lung cancer, chromosomal abnormalities were predominantly polyploid (the most common are triploid and pentaploid), and only a few are haploid(Fig. 3B and 3D) (due to the very small number of haploids, there was no statistical subtype), which was in line with the literature reports that the polyploid genetic material was relatively abundant, conducive to the development of malignant biological behavior of cells. the vast majority of CTCs in patients with benign nodules were triploid, but in patients with cancer, especially in patients with advanced stages (Fig. 3B and 3D), the pentaploid begun to increase, explaining that the increase in genetic material was likely to increase the degree of malignancy. Although CTECs were also dominated by triploid and pentaploid, pentaploid account for the vast majority of patients, especially in advanced patients (Fig. 3C and 3E).
Distribution of CTC and CTEC with different size in patients
Peripheral blood CTCs and CTECs vary in size. The definition of large and small cells is generally based on white blood cells (5um), cells larger than 5um are defined as large cells, and cells smaller than 5 cm are defined as small cells, and the standard image is shown in Fig. 4A. The biological and clinical significance of CTCs are different. For example, small CTCs has been reported to be significantly associated with recurrence and disease-free survival (DFS) of liver cancer[10]. However, the correlation between the size of CTECs and its clinical significance is still unclear. Therefore, we separately calculated the distribution of size of CTCs and CTECs in patients with different stages of cancer. On average, except for small CTECs, both large and small CTCs and large CTECs had the effect of distinguishing between benign and malignant nodules and staging (Fig. 4B-4E). But compared to other subtypes, large CTCs had an advantage in identifying early and late lungs(Fig. 4B) (P < 0.0001), while large CTECs were more able to distinguish between benign nodules and early stage lung cancer patients༈Fig. 4D) (P = 0.0085). In comparison, small CTECs were not only small in number, but also cannot do good and bad in prediction and staging diagnosis (Fig. 4E). But when we compared the proportion of large CTCs and large CTECs in different patient populations, we found that there was no difference in the proportion of total circulating cells in the two sub-categories, especially large CTCs, in four different populations (Fig. 4F)
Chromosome ploidy distribution of different size of CTCs and CTECs
Since chromosomal ploidy of CTCs and CTECs subclasses with different sizes have certain clinical significance, we next analyzed the chromosomal ploidy differences of large and small cells. We observed a significant difference in the cell size of different ploidy numbers (Fig. 5A-5D). According to the standard of 5um in diameter, CTCs were divided into large cells and small cells, and small cells were mainly triploid, while large cells mostly were pentaploid. The number of tetraploid was between that of triploid and pentaploid, and its size (large or small) was not much different (Fig. 5A and 5B), but the proportion of tetraploid in large CTECs was significantly higher than that in small CTECs (Fig. 5C and 5D). However, as shown in Fig. 4, the large cells were more significant than small cells in staging diagnosis between normal humans, benign nodules, and lung cancer.
Subclass analysis of CTCs and CTECs based on chromosomal ploidy and cell size combination
To further understand the significance of the various subtypes of CTCs and CTECs in the early diagnosis and staging of lung cancer, we divided CTCs and CTECs respectively into 24 subgroups based on cell size, chromosome ploidy and four stages (normal, benign, early, and late) of tumor development for separate analysis (Fig. 6A-6D).
Among the large CTCs, all the triploid, tetraploid and pentaploid could distinguish between early and late patients(P < 0.05), but only the pentaploid could identify benign nodules and early patients (Fig. 6A). And compared to the other two ploidy, the ability of the pentaploid to distinguish between early and late patients was stronger(P < 0.001). For small CTCs, tetraploid had obvious early diagnostic significance(P < 0.001), while pentaploid had no advantage in both early diagnosis and staged diagnosis. However, because tetraploids accounted for only a small proportion of the entire CTC population, it might affect the sensitivity of its clinical application. When we combined the large CTCs and the small CTCs to reduce the subgroups, the sub-categories obtained after the analysis were similar in the distribution of the large CTCs in the different stages (Fig. 6A and 6E). Current reports on CTC ploidy tend to combine tetraploid and pentaploid into a subgroup, collectively referred to as polyploid. Compared to triploids, polyploids do find some clinical significance, such as related to metastasis, recurrence and drug resistance[9]. Therefore, we also tried to combine triploid and tetraploid into a subclass, namely, polyploid, and then compare the significance of this new subclass in the diagnosis of tumor staging. It was found that for large CTCs, the polyploid subclass after the combination had no obvious advantage in distinguishing between benign and malignant nodules and distinguishing between early and late patients compared with the single subclass before the merger (Fig. S1A). Moreover, once merged, the ability of small CTCs in identifying benign and malignant nodules was lower than before (Fig. S1B).
For CTECs, we did the same analysis. Large pentaploid CTECs could simultaneously identify benign and malignant nodules and classify for patients with early and late stage tumors (Fig. 6B). Small triploid CTECs could distinguish between early and late patients, but could not identify benign nodules and early patients (Fig. 6D). But after the combination of large and small CTECs, both tetraploid and pentaploid had certain clinical significance. The pentaploid was only meaningful for early and late stage patients, and tetraploid could distinguish between benign nodules and early lung cancer patients (Fig. 6F). When we combined the tetraploid and pentaploid of CTECs into a polyploid subclass, small polyploid CTECs could not stage clinical patients (Fig. S1D). And large CTECs had the same effect as before the merger, that is, only pentaploid was meaningful, and could distinguish between benign nodules and early patient or between early and late patients (Fig. S1C).
Based on the results of CTCs and CTECs, we found that for CTCs, the most advantageous one for identifying benign nodules and early patients was tetraploid small CTCs; the most appropriate for identifying early and late patients was the pentaploid large CTCs; For CTECs, the sub-class of best identifying benign nodules and early patients was the combination of large and small tetraploid cells or large polyploid cells. But the best choice for distinguishing between early and late patients was large pentaploid CTECs or large polyploid (tetraploid and pentaploid) CTECs. Finally, comparing CTCs and CTECs as a whole, whether in differentiating of benign and malignant nodules or of early and late lung cancer, CTCs were slightly more advantageous than CTECs.
CTEC was not a senescent or apoptotic cell, but an active cell population
In principle, endothelial cells are non-tumor cells, and the genetic material is relatively stable, while CTEC chromosomes exhibit different degrees of copy number variation, similar to the malignant transformation process of tumor cells. However, because these cells are strongly positive for CD31, although they have many similarities with tumor cells, they are classified as a column of endothelial cells. We suspected that these CTECs that entered the blood were all aging or impending apoptosis, therefore they showed chromosomal variation. However, we had discovered some new phenomena, such as the phenomenon of cell division in CTECs isolated from peripheral blood (Fig. 7A), and CTECs could be clustered like tumor cells(Fig. 7B) or grouping with other types of cells such as white blood cells(Fig. 7C) and tumor cells, in line with the literature reports[11]. The binding of CTCs to leukocytes can promote their own malignant phenotype[12], but the clinical significance of CTECs and leukocyte or tumor cell agglomeration is still unclear. At the same time, some CTECs could also express tumor markers such as vimentin (Fig. 7D). These phenomena together indicated that aneuploid CTECs could indeed exist as a separate active cell population like a tumor cell, rather than simply shed cells that fall into the blood after aging or death. But whether CTECs have potential malignant behavior and its true biological differences with CTCs requires further cellular and molecular experiments to validate, and the future will help to understand these issues with single-cell sequencing.