Lymph Liquid Biopsy for Detection of Cancer Stem Cells

Abstract

Collection of a blood sample defined by the term “blood liquid biopsy” is commonly used to detect diagnostic, prognostic, and therapeutic decision-making markers of metastatic tumors including circulating tumor cells (CTCs). Many tumors also release CTCs and other markers into lymph fluid, but the utility of lymphatic markers largely remains unexplored. Here, we introduce lymph liquid biopsy through collection of peripheral (afferent) and central (thoracic duct [TD]) lymph samples and demonstrates its feasibility for detection of stem-like CTCs potentially responsible for metastasis development and tumor relapse. Stemness of lymphatic CTCs (L-CTCs) was determined by spheroid-forming assay in vitro. Simultaneously, we tested blood CTCs by conventional blood liquid biopsy, and monitored the primary tumor size, early metastasis in a sentinel lymph node (SLN) and distant metastasis in lungs. Using a mouse model at early melanoma stage with no distant metastasis, we identified stem-like L-CTCs in lymph samples from afferent lymphatic vessels. Since these vessels transport cells from the primary tumor to SLN, our finding emphasizes the significance of the lymphatic pathway in development of SLN metastasis. Surprisingly, in pre-metastatic disease, stem-like L-CTCs were detected in lymph samples from the TD, which directly empties lymph into blood circulation. This suggests a new contribution of the lymphatic system to initiation of distant metastasis. Integration of lymph and blood liquid biopsies demonstrated that all mice with early melanoma had stem-like CTCs in at least one of three samples (afferent lymph, TD lymph, and blood). At the stage of distant metastasis, spheroid-forming L-CTCs were detected in TD lymph, but not in afferent lymph. Altogether, our results demonstrated that lymph liquid biopsy and testing L-CTCs holds promise for diagnosis and prognosis of early metastasis.

Liquid biopsy is an advanced emerging method in clinical and experimental oncology that refers to detecting circulating tumor markers in blood samples (1–6). Among others (e.g., tumor-associated exosomes, RNA, and DNA), circulating tumor cells (CTCs) are direct “messengers” and exclusive initiators of deadly tumor metastasis (6–10). To date, multiple reports have demonstrated that blood liquid biopsy is a useful method to diagnose subpopulations of blood CTCs (B-CTCs), including cancer stem cells, that can initiate metastasis and tumor relapse (11–13). However, despite significant efforts, blood liquid biopsy still has limited success, and many patients continue to die of metastasis (6,14).

One of the critical barriers in blood liquid biopsy is that CTCs can escape from blood circulation by taking the lymphatic pathway for dissemination. In contrast to blood, which circulates through the entire body, lymph vascularity is characterized by one-way flow (15). The afferent peripheral lymph is formed in lymphatic capillaries through capturing fluid, cells, and other substrates from surrounding tissues and transporting them by afferent lymphatic vessels in the proximal direction to the nearest (sentinel) lymph node (SLN), which is likely the first metastatic site (15–16). According to numerous reports, many early primary tumors (e.g., melanoma) are able to shed CTCs into afferent lymphatic vessels to deliver them to the SLN (17–22). In SLNs, lymphatic CTCs (L-CTCs) can initiate metastasis, which is traditionally considered as a manifestation of metastatic disease (23–24). According to recent data, L-CTCs may also transit the SLN and enter the blood circulation for systemic dissemination (25). Otherwise, L-CTCs that have bypassed the SLN might continue transportation through the network of lymphatic vessels and lymph nodes to reach the thoracic duct (TD), which empties lymph into one of the main veins near the heart (15). This leads to the intriguing hypothesis that there is a significant direct role of the lymphatic system in initiation of distant metastasis.

Despite the obvious importance of L-CTCs, we have only fragmentary knowledge on the contribution of L-CTCs to formation of metastases, including a few reports that described the occurrence of L-CTCs in afferent vessels (26–27). We do not know yet whether L-CTCs can be disseminated by the TD, and whether lymph can transport cancer stem cells. Furthermore, from the aforementioned considerations, afferent peripheral lymph and TD central lymph samples might have different diagnostic values that have never been highlighted.

To fill these gaps, we introduced a concept of lymph liquid biopsy of TD and afferent lymph for monitoring tumor markers (Fig. 1). Here, we used a mouse model of melanoma, which disseminates CTCs through the lymphatic system; it has a high metastasis-related mortality and, thus, demands new diagnostic tests to identify early and potentially treatable stages of disease (14,19,23–24,28). We focused our study on lymph liquid biopsy from afferent lymphatic vessels and TD in pre-metastatic and early metastatic melanoma. As metastases have been suggested to initiate from a subpopulation of cancer (tumor) stem cells, we explored the existence of these cells among bulk L-CTCs using the well-established and distinguishable feature of stem cells to form spheroids over culturing in vitro.

Fig 1.

Principle of lymph liquid biopsy of afferent and TD lymph.

Materials and Methods

Tumor Cell Lines

Mouse melanoma cells (B16-F10 cell line, American Type Culture Collection, Manassas, VA, USA) were cultured per the vendor’s specifications. The viability of the cells was checked with the Trypan Blue test.

Mouse Model

In vivo experiments were done with mice in accordance with protocols approved by the UAMS Institutional Animal Care and Use Committee. Mice (nu/nu; weighing 20–25 g) were purchased from a commercial source. To prevent undue suffering, mice were used only within an ethical framework based on the 3Rs principle.

To establish a primary tumor, mice were inoculated in an ear with 0.5 × 10⁶ melanoma cells in ~5 μl of phosphate buffer solution (PBS). Growth of the primary tumor was visible as a black spot by the naked eye that allowed us to monitor and measure its size in vivo using calipers. The mice were anesthetized with isoflurane when it was needed.

Evans Blue Lymphography of Afferent Lymphatic Vessels

For lymphography, the anesthetized animal was positioned on its back on the heated microscope stage (at 37.7°C, body temperature). To visualize optically transparent lymphatic vessels, we used a well-established procedure of injection of Evans Blue (EB) dye (1–3 μl of 1% solution; MP Biomedicals, Inc., Santa Ana, CA, USA) around the primary tumor. A few minutes after injection of the EB dye, afferent ear lymphatic vessels became visible as blue channels between the primary tumor and the SLN.

Collection of Afferent Lymph

After EB lymphography, the skin under the afferent lymphatic vessel was incised to prepare a skin flap exposing the blue vessels ending by the SLN (cervical lymph nodes for ear tumors). An afferent lymphatic vessel was carefully separated from the surrounding tissue to 1 cm in length using microsurgical scissors and forceps.

To harvest lymph from the freshly dissected lymphatic vessel, it was first ligated in its proximal part. In 20–30 min, the distal part of the vessel was enlarged due to lymph accumulation. At this time, the distal part of the vessel was ligated, and a fragment of vessel with lymph was extracted from the organism and immediately placed into the medium-filled petri dish. Here, the ligatures were removed to allow the lymph fluid and cells to empty from the vessel into the medium. After that, L-CTCs were imaged and tested in vitro.

To sample lymph by cannulation of afferent lymphatic vessels, a glass microcannula (custom made in our laboratory), connected with a syringe, was slowly inserted into the vessel. A magnifying glass (F378650000/EMD, Bel-Art, Wayne, NJ, USA) was used for successful cannulation. Lymphatic cannulation for ~30–40 min had a yield of a few microliters of pure afferent lymph.

Sampling TD Lymph

First, 1 ml of milk cream (7.5% fat) was administered into the mice using oral gavage 30 min before TD lymph sampling. For oral gavage, we used a plastic feeding needle (flexible, disposable), 20 ga × 38 mm (FTP-20–38, Instech, Plymounth Meeting, PA, USA). Then, the animal was positioned on its right side on the heated microscope stage. The abdomen was opened through a left subcostal incision and extended 1.5 cm laterally. Using a self-retaining retractor, the organs were immobilized. Using microsurgical scissors and forceps, the dorsal parietal peritoneum was incised above the adrenal gland, and the aorta was cleared from the surrounding fatty tissue, exposing the cisterna chyli and the TD. Approximately 30 min after the gavage, both of them were easily visualized as white-colored structures due to the ingested cream. A glass cannula with heparinized saline (50 IU/ml) was inserted into the TD, and ~100 μl of TD lymph was harvested during several minutes.

Blood Collection

Following the lymph liquid biopsy, mouse blood was collected to study B-CTCs using cardiac puncture with a plastic sterile syringe containing anticoagulant. To examine B-CTCs, immediately after sampling, erythrocytes were removed using lysis buffer (Miltenyi Biotec, Bergisch Gladbach, Germany). The residual cell suspension containing B-CTCs was used for in vitro tests.

Optical Microscopy

A cooled, color CCD camera (DP72, Olympus-NDT, Waltham, MA, USA) was operated in transmission mode at different magnifications (from ×10 to ×100, oil immersion) and was used to image lymph and blood samples (1) on the slide immediately after sampling ex vivo and, then, (2) in the wells over the course of 3 weeks for monitoring growth of spheroids in the plates in vitro.

Spheroid Formation Assay

Cells from three types of samples (i.e., afferent lymph, TD lymph and blood) were seeded in the wells (1–3 cells/well) of a 96-well plate and cultured in an incubator for 3 weeks. Wells were filled by DMEM/F12 with L-glutamine (Corning Inc., Corning, NY, USA) with 10 ng/ml of rmFGF-basics (R&D systems, Minneapolis, MN, USA), 20 ng/ml of rmEGF (R&D systems, Minneapolis, MN, USA), 0.05 mg/ml of insulin solution from bovine pancreas (Sigma-Aldrich, St. Louis, MO, USA). The medium was carefully changed two times a week. Melanoma spheroids were counted weekly using microscopy. The percentage of wells with spheroids was used to quantify the spheroid-forming activity.

Analysis of CD45+ Cells

To identify leukocytes, cells in intact lymph samples extracted from healthy mice were stained with 0.2 mg/ml of APC anti-mouse CD45 monoclonal antibody using the protocol provided by the company (eBioscience, San Diego, CA, USA). Then, samples were fixed with 300 μl of 2% solution of formaldehyde in PBS for 10 min at 4°C. Finally, the cells were analyzed by flow cytometry (LSRFortessa, BD Biosciences, Franklin Lakes, NJ, USA).

Histology and Immunohistochemistry

After liquid biopsies, the primary tumor, SLN, bilateral lungs, heart, liver, and brain were removed. Each tissue was placed in 10% neutral buffered formalin for fixation, embedded in paraffin, and cut into histologic sections. The 5–7 mm thick histologic sections were stained with hematoxylin and eosin (H&E) and evaluated by light microscopy to determine metastasis.

Statistical Analysis

Data for each experiment were summarized by time-point group as the mean ± SEM (standard error of the mean). To provide sufficient material for statistical analysis and ensure robust and unbiased results, we used ≥3 animals per group. The differences were considered statistically significant if P ≤ 0.05.

Results and Discussion

To produce metastatic melanoma, we inoculated B16-F10 melanoma cells into the ears of nude mice (one ear per mouse). In pursuit of defining the role of L-CTCs in initiation and development of early metastasis, we first defined pre-metastatic and early SLN metastatic stages of disease. From our experience with this model, the mice do not have any histological signs of metastasis or show early and small SLN metastasis at Week 1 after inoculation (29–30).

In the current study, all mice at Week 1 after inoculation had a well-defined primary tumor (Fig. 2a,b). Its size was measured with calipers and tumor volume was calculated as ½ × long diameter × short diameter² (30–31). The tumor volumes had an average ± SEM of 8.9 ± 2.67 mm³. At this stage, H&E histological examination showed that approximately 1/3 of mice had no signs of any metastasis. The other 2/3 of mice exhibited a small metastasis in the SLN (Fig. 2c–e). On average, metastatic lesions occupied 6.6 ± 2.41% of an SLN’s volume (Fig. 2c). We did not find histologically detectable metastases in distant organs. At this time point, melanoma-bearing mice were the subjects of lymph liquid biopsy to collect lymph samples and explore spheroid-formation potential of L-CTCs in vitro.

Fig 2.

Mouse model of metastatic melanoma. (a) Diagram of the experiment (left) and intravital images (right) of primary tumor (top right) and blue-colored afferent lymphatic vessel (LV) after EB lymphography (bottom right). (b) Individual (diamond plot) and average volume of primary tumor (column, mean ± SEM, n = 11) in the group of mice at week 1 after melanoma inoculation. (c) Percentage of early SLN involvement. (d) H&E histology of intact SLN. (e) Histologically (H&E) detectable early SLN metastasis (left) and the enlarged fragment of the metastatic nodule with melanoma-specific brown pigment melanin (right).

Through extensive search, we were able to find only a few reports about sampling lymph in animals and humans (32–35). In most of them, lymph was collected using direct cannulation of lymphatic vessels and TD (32–34). Recently, a few other methods have been reported for detecting lymphatic extracellular tumor-related vesicles that include collecting lymph-containing exudative fluid from around the tumor and SLN, and harvesting lymph from freshly dissected lymphatic vessels (35–36).

In our study, we initially tried lymph liquid biopsy through sampling lymph from freshly dissected afferent lymphatic vessels that drain from the primary tumor. For this, afferent lymphatic vessels in melanoma-bearing mice (n = 5) at Week 1 post-inoculation were visualized with EB lymphography (Fig. 2a, right and bottom). The dissected vessel fragment with lymph fluid was emptied into the petri dish with medium for growing melanoma stem cells. Microscopic imaging of fresh samples confirmed the presence of lymphatic cells. However, culturing cells in the incubator revealed that samples became heavily contaminated by growing endothelial-like cells in a few days, which prevented monitoring of spheroids later. Thus, while this approach is acceptable to detect extracellular microparticles in lymph samples (35), it was not well-suited to study L-CTCs. There are two ways to overcome this methodological problem: (1) by collecting lymph through direct cannulation of the lymphatic vessel inside the body; or (2) by extending the protocol with dissected vessels using additional procedure(s) to separate L-CTCs from endothelial and other cells. However, assuming the extreme rarity of L-CTCs, especially stem L-CTCs at early stages of disease, any additional procedure with a lymph sample significantly elevates the risk of losing these cells and, eventually, decreases diagnostic value. Because of these considerations, we decided to implement lymph liquid biopsy through cannulation.

In this set of experiments, mice were the subjects of lymph liquid biopsy through sequential cannulations of (1) afferent lymphatic vessel between a primary tumor and its SLN (the cervical lymph node for ear tumors) and (2) TD (n = 11, Fig. 3a,b). The lymph liquid biopsy was integrated with a test for B-CTCs through conventional blood liquid biopsy that included blood collection and removal of erythrocytes to get a sample of leukocytes and CTCs. Thus, we received fresh samples of three fluids (i.e., afferent lymph, TD lymph and blood) and tested them using the same protocol. Approximately 8 μl of each sample was imaged on the slide with a 120-μm-thick well by optical microscopy at different magnifications (Fig. 3c). The rest of the sample was put into wells of a 96-well plate (1–3 cells/well) and cultured in vitro. Imaging the wells over 3 weeks allowed us to calculate the percentage of wells that formed spheroids. By this design, we were able to test the spheroid-forming potential of both afferent and TD L-CTCs as well as blood CTCs in the same mouse at the same time point of disease, and correlate occurrence of stem-like CTCs with size of primary tumor and status of metastasis.

Fig 3.

Sampling lymph using cannulation of lymphatic vessels. (a) Cannula (C) in afferent lymphatic vessel (LV). (b) TD after 30 min of milk cream gavage in vivo before cannulation. (c) Optical images of intact (i.e., from healthy mouse) TD lymph sample on the slide in vitro obtained at 10× magnification of the objective (right) and 60× magnification of the objective (oil immersion). (d) Identification of CD45+ cells from the total population of TD lymphatic cells using flow cytometry; the representative experiment is shown.

To make sure that normal lymphatic cells do not grow in spheroids over 3 weeks of culturing, we used a control group of healthy mice (n = 5). In these mice, lymph samples from afferent ear vessels and TD were collected by cannulation. Imaging of afferent lymph demonstrated relatively low cell concentration. Most cells were transparent and round-shaped suggesting that these cells were leukocytes. In some samples we also found rare erythrocytes that were recognized by their biconcave specific shape. As expected (15), TD lymph samples had a significantly higher concentration of cells, many of them being leukocytes and erythrocytes (Fig. 3c). To confirm the phenotype of lymphatic cells, we analyzed TD lymph samples (n = 3) by flow cytometry. Leukocytes were defined as CD45+ cells (Fig. 3d). We observed that normal lymphatic cells had a significant proportion of CD45+ cells that varied between 44.1% and 53.2%. Culturing and periodic imaging of lymphatic cells in wells with the same media used to grow melanoma spheroids showed that normal lymphatic cells died or disintegrated in a few days. Obviously, there was not any spheroid over 3 weeks of monitoring.

In contrast to intact lymph, imaging of the microscopic slides with lymph samples from afferent lymphatic vessels of mice with early melanoma showed rare black-pigmented cells with a larger size than leukocytes (Fig. 4a). These “black” cells were associated with melanoma cells. Given time, some CTCs proliferated and formed well-defined black-colored spheroids (Fig. 4b). Specifically, culturing in vitro for up to 3 weeks demonstrated that lymph samples from 45% of mice had spheroid-forming L-CTCs. These data showed that early primary melanoma tumors released stem-like tumor cells into lymph, which transports L-CTCs by afferent lymphatic vessels to the SLN. We did not find any statistically significant difference in spheroid-forming activity when we compared SLN-negative and SLN-positive groups of mice (at the time of liquid biopsy). Then we analyzed the relationship between L-CTCs and primary tumor volume. It was interesting that rare “black” L-CTCs were found in lymph samples of mice with a very small primary tumor, sized around 1–1.5 mm³, but L-CTCs from these mice did not grow in spheroids. Measuring primary tumors in groups of mice with (n = 5) and without (n = 6) stem-like L-CTCs demonstrated that stemness was the attribute of primary tumors with significantly bigger sizes (averages ± SEMs of 15.0 ± 4.02 mm³ with stemness vs. 3.80 ± 1.99 mm³ without stemness; P = 0.027) (Fig. 4c). These findings suggest that primary tumors should achieve a certain size to release stem L-CTCs.

Fig 4.

Lymph liquid biopsy of stem L-CTCs. (a) High-resolution optical images (60× magnification of the objective, oil immersion) of individual L-CTC in the lymph sample on the slide immediately after lymph liquid biopsy. (b) Incidence (% mice and % wells on the plate; each plate contained cells from the same afferent lymph sample of one mouse) of growing melanoma spheroids from afferent L-CTCs sampled from the mice at Week 1 post-inoculation; insert: optical image (20× magnification of the objective) of the afferent L-CTC originated spheroid. (c) Difference in primary tumor sizes between group of mice (n = 5), which had afferent L-CTC originated spheroids and the group of mice, which afferent L-CTCs did not form spheroids; diamond plots are individual results, black columns is mean ± SEM. (d) Incidence (% mice and % wells on the plate) of growing melanoma spheroids from TD L-CTCs sampled from the mice at Week 1 post-inoculation; insert: optical image (20× magnification of the objective) of the TD L-CTC originated spheroid. (e) Incidence (% mice) of growing melanoma spheroids from B-CTCs sampled from the mice at Weeks 1 and 3 post-inoculation. (f) Incidence (% mice) of growing melanoma spheroids from afferent and TD L-CTCs sampled from the mice at Week 3 post-inoculation.

Unexpectedly, L-CTCs were found in TD lymph samples (Fig. 4d). Many TD samples were able to form spheroids, including some from mice without any metastasis. This means that the TD can deliver stem-like L-CTCs into the central blood circulation toward distant organs before the development of histologically detectable metastasis. It is currently unclear how L-CTCs appear in the TD, but we can speculate that L-CTCs might transit SLN to continue traveling by using either lymphatic or hematogenous CTC pathways as a result of cell migration between blood and lymphatic systems (21).

Our testing of blood samples demonstrated the presence of spheroid-forming “black” B-CTCs (Fig. 4e) assuming parallel dissemination of stem CTCs by lymphatic and blood systems.

Comparative analysis of spheroid-forming activity in afferent lymph, TD lymph, and blood showed that all tumor-bearing mice had spheroid-forming stem-like CTCs in at least one of the three fluids. However, three fluids were involved in stem-like CTC dissemination in 18% of mice only, and these mice already had SLN metastasis. In the majority of mice (82%), spheroids were detected in two (afferent lymph + TD lymph, afferent lymph + blood, or TD lymph + blood) or one fluid. Furthermore, we determined that 36% of mice exhibited stem-like CTCs in lymph fluids, not in blood. Thus, our results can explain false negative results of blood liquid biopsy, and indicate that examination only of blood samples is not enough for prognosis of metastatic disease.

Finally, we tested mice at 3 weeks post-inoculation. At this stage, all mice had advanced SLN metastasis with initial distant lung metastases. In contrast to early-stage samples, L-CTCs from late-stage afferent lymph did not form spheroids, whereas TD L-CTCs (67%) and some B-CTCs continued to maintain spheroid-forming activity (Fig. 4e,f). The stopping of L-CTC release into afferent lymph might be explained by evolution of primary tumors, the growth of which is typically associated with increased proportions of apoptotic and necrotic cells (37).

Conclusions

Combined lymph liquid biopsy through cannulation of lymphatic vessels and TDs was developed for testing spheroid-forming stem-like L-CTCs in peripheral and central lymph fluid sampled from mice with pre-metastatic and early metastatic melanoma. Using this approach, we demonstrated that early primary tumors are able to shed stem-like L-CTCs in afferent lymph that transport cells to the SLN. Finding stem-like L-CTCs in TD lymph samples evidences that TD in early melanoma can deliver viable and stem-like CTCs into the venous system, bypassing narrow capillaries. These experimental data suggest to reconsider the role of lymphatic system toward its direct involvement in an initiation of distant metastasis.

Our results indicate the importance of integrated lymph and blood liquid biopsies, because CTCs can use both the lymphatic and blood pathways for dissemination in a currently unpredictable manner. The benefits of integrated liquid biopsy include obtaining complementary information and increasing the diagnostic accuracy in CTC diagnosis.

Altogether, these experimental results evidenced the existence of a “lymphovascular” niche for dissemination of stem-like CTCs, which is in line with a previously published hypothesis (22), and suggest development of clinically relevant approaches of lymph liquid biopsy. The exciting perspectives are based on using patient-derived animal models and a recently developed radiological method to sample the TD lymph in humans (38–39).

Further extension of using lymph liquid biopsy to detect other markers (e.g., circulating tumor-related DNA and miRNA) in metastatic tumors would be the foundation for discovering lymphatic-related mechanisms of metastasis and development of novel tumor biomarkers.