Human NSCLC cell lines (H358, HCC827, H1993) and human normal fibroblasts (MRC-5) were used in this study and purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA). H358, HCC827, H1993 and MRC-5 cells were cultured in RPMI-1640 medium Gibco^® (Thermo Fisher Scientific, Inc., Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS; Gibco^®; Thermo Fisher Scientific, Inc.) in a humidified atmosphere of 5% CO2 at 37°C. All cell cultures underwent regular testing for Mycoplasma contamination and consistently yielded negative results.

Before co-culture, over 80% cell viability was confirmed by Trypan blue staining. H358, HCC827 or H1993 and MRC-5 cells were counted and seeded at a ratio of 1:3-1:4 for direct co-culture in 150 cm² flasks in RPMI-1640 medium Gibco^® supplemented with 10% FBS in a humidified atmosphere of 5% CO2 at 37°C (25). The medium was changed every other day. Trypsin was used to adjust the percentage of co-cultured cells. After adding trypsin and incubation at 37°C for about 1 minute, most MRC-5 cells became non-adherent and were washed with PBS for the next step of co-culture.

After 14-day direct co-culture, cells were seeded on sterilized cover slips in 12-well plates in a humidified atmosphere of 5% CO2 at 37°C. Upon reaching 70-80% confluence in a single layer, cells were fixed by 4% paraformaldehyde and permeabilized by 0.5% Triton X-100 in PBS (PBST) (25). Cells were blocked in blocking buffer (3% bovine serum albumin in PBST) at room temperature for 30 minutes. Primary antibodies (α-SMA, 1A4, ab7817, abcam, 1:200; α-SMA, D4K9N, #19245, Cell Signaling Technology, Inc., 1:200; CK7, D1E4, #4465, Cell Signaling Technology, Inc., 1:500; VEGF-C, E-6, sc-374628, Santa Cruz, 1:200; VEGF-D, C-12, sc-373866, Santa Cruz, 1:50; PDGFRβ, D-6, sc-374573, Santa Cruz, 1:200; PDPN, 18H5,sc-59347,Santa Cruz, 1:200); were diluted to in blocking buffer and incubated at 4°C overnight. Alexa Fluro™ 568 donkey anti-mouse IgG (H+L) (#A-10037, Invitrogen; Thermo Fisher Scientific, Inc., 1:500) and Alexa Fluro™ 488 goat anti-rabbit IgG (#A-10034, Invitrogen; Thermo Fisher Scientific, Inc., 1:500) were diluted in blocking buffer and incubated at 4°C in the dark for 1 hour. Prolong^® Gold anti-fade reagent (Invitrogen; Thermo Fisher Scientific, Inc.) containing 4’,6-diamidino-2-phenylindole (DAPI) was added to the cover slips. Slides were sealed with nail polish. Images were captured by a fluorescence microscope (Nikon, Tokyo, Japan).

Cells were collected and lysed in sodium dodecyl sulphate (SDS) lysis buffer (1% SDS, 10 mM HEPES, pH7.0, 2 mM MgCl2, universal nuclease 20 U/ml). Total cellular protein concentration was measured using a BCA assay kit (Thermo Fisher Scientific Inc.). Equal amounts of the protein samples were subjected to 10% SDS-PAGE and transferred to nitrocellulose blotting membranes (GE Healthcare). The membranes were then blotted with primary antibodies α-SMA, [1A4, ab7817, abcam, 1:200]; Col 1A1[E3E1X, #66948, Cell Signaling Technology, Inc., 1:1000]; GAPDH [sc-47724, Santa Cruz Biotechnology], 1:1000) at 4°C overnight, then incubated with the secondary antibody (horse anti-mouse IgG, HRP-linked antibody (#7076, Cell Signaling Technology, Inc.,1:1000); goat anti-rabbit IgG, HRP-linked antibody (#7074, Cell Signaling Technology, Inc.,1:1000)) for 1h at room temperature. Proteins were visualized using enhanced chemiluminescence exposed on autoradiograph film and developed using standard methods previously described (26).

After identification of α-SMA⁺ fibroblasts by immunofluorescence, co-cultured cells were harvested in ice cold phosphate buffered saline (PBS) and around 5×10⁵ cells allocated per tube. To stain cancer cells, PE anti-human CD326 (EpCAM) antibody (9C4, #324206, Biolegend, 1:200) and PE mouse IgG2b, Isotype Ctrl antibody (MPC-11, #400312, Biolegend, 1:500) were diluted in ice cold fluorescence-activated cell sorting (FACS) buffer and incubated on ice for 30 minutes in the dark. After washing with FACS buffer, cells were fixed and permeabilized by buffer (4% paraformaldehyde and 0.5% Triton X-100 in PBS) on ice for 30 minutes in the dark. To stain α-SMA⁺ fibroblasts, anti-α-SMA antibody (1A4, ab7817, abcam, 1:200) was diluted in washing buffer and incubated at 4°C for 1 hour in the dark. After washing with FACS buffer, goat anti-mouse IgG-FITC (sc-2010, Santa Cruz Biotechnology, 1:500) was diluted in washing buffer and incubated on ice for 30 minutes in the dark. Cells were then acquired on CytoFLEX (Beckman Coulter, Inc., USA) for flow cytometry analysis. Data analysis was performed using FlowJo™ (BD biosciences).

Cell sorting was performed as described previously (25). EpCAM was utilized as a surface marker to isolate H358 or HCC827 cells from MRC-5 cells. After 14 days of co-culture, cells were incubated with PE anti-human CD326 (EpCAM) antibody (9C4, #324206, Biolegend, 1:200). PE mouse IgG2b, Isotype Ctrl antibody (MPC-11, #400312, Biolegen, 1:500) was added to cells as a control for gating during cell sorting. EpCAM⁺ and EpCAM^- cells were isolated by MoFlo XDP (Beckman Coulter). Data analysis was also performed on FlowJo™.

Female BALB/cAnN-nu (nude) mice (4–6 weeks) were used for intrathoracic injections. After acclimatization for one day, mice were anesthetized by intraperitoneal administration of ketamine (100 mg/kg) and xylazine (10 mg/kg). H358 or HCC827 cells (2×10⁶) with or without lung CAFs (0.5×10⁶) were re-suspended with growth factor reduced Matrigel^® (Corning) in PBS (total volume < 40 μl) using insulin syringes (BD) and kept on ice. Subsequently, cells were directly injected into the left lung of each mouse (15). The technique of intrathoracic injection is described in greater detail below. After confirming sufficient anesthesia by toe pinch, the mouse was placed in a lateral decubitus position on a warm blanket. Skin was disinfected with 10% betadine solution, then wiped off using 70% alcohol cotton swabs. The left scapula was located by gentle determination of the subscapular angle. Then along the subscapular angle, the upper and lower ribs were felt with the ventral side of the left index finger. The point at which the perpendicular line of the subscapular angle intersected the horizontal line of the ribs was the injection site. The needle was advanced rapidly to a depth of approximately 5 mm into the left lobe of the lung between the 6th and 7th ribs, the plunger withdrawn to exclude puncture of any blood vessels and the cell suspension injected slowly. After complete injection, the needle was maintained in the lung for 15 seconds to avoid leakage of the injected material, then quickly removed and pressure applied to the area with cotton swabs soaked with 70% alcohol until hemostasias was achieved. The mouse was monitored closely until full recovery from anesthesia. All mice experiments were approved by the Committee on the Use of Live Animals in Teaching and Research, The University of Hong Kong (CULATR 5580-20).

Before performing micro-CT (Bruker, SkyScan1276) scan, mice were deeply anesthetized by intraperitoneal injection of ketamine (100 mg/kg) and xylazine (10 mg/kg). Mice were then placed prone in the micro-CT chamber and wrapped in a bed using paper adhesive tape to avoid any movement (27). The micro-CT scan time was adjusted to 45 seconds per mouse. After scanning, mice were placed in a warm incubator to recover. Micro-CT scan was performed weekly to monitor tumor formation and progression. Images were analyzed by Dataviewer (Bruker) and Fiji (ImageJ, NIH). Quantification of tumor size was based on the Response Evaluation Criteria in Solid Tumor (RECIST 1.1) (28).

Mice were euthanized with intraperitoneal injection of pentobarbital. A lower abdominal incision was made to expose the organs and to detect metastasis. Mice were exsanguinated by cutting the abdominal aorta and inferior vena cava. Diaphragm was carefully removed to detect primary tumor, pleural effusions, and intrathoracic metastasis. Then rib cage with sternum was cut off to expose the heart, lungs, and mediastinum. 25G needle attached with a syringe containing 5 ml cold PBS was inserted into the right ventricle of the heart to perfuse the lung tissue. After removal of heart, the whole mediastinum with lungs were resected and washed in cold PBS. The samples were immediately fixed in 4% paraformaldehyde for 24–48 hours followed by embedding in paraffin (15).

Formalin-fixed paraffin embedded (FFPE) tissues were cut at 4-6 μm, dewaxed in xylene, and rehydrated through graded ethanol solutions. For hematoxylin & eosin (H&E) staining, slides were immersed into hematoxylin solution for 3–5 minutes and eosin solution for 1 minute. Primary tumor and mediastinal lymph node metastasis were examined under a microscope. For IHC staining, following deparaffinization and rehydration, slides were boiled in microwave for antigen retrieval (10 mM sodium citrate, 0.05% Tween 20, pH 6). Then, slides were immersed into blocking buffer (BLOXALL^® SP-6000-100, Vector Laboratories, Inc.) for at least 15 minutes to block endogenous peroxidase and alkaline phosphatase. After permeabilization (0.5% Triton X-100 in PBS) and non-specific binding blocking (2.5% normal horse serum, MP-7500-50, Vector Laboratories, Inc.), slides were stained with anti-CK7 (D1E4, #4465, Cell Signaling Technology Inc. 1:500; VEGF-C, E-6, sc-374628, Santa Cruz, 1:200; VEGF-D, C-12, sc-373866, Santa Cruz, 1:50);, anti-Ki-67 (8D5, #9449, Cell Signaling Technology Inc. 1:500), anti-α-SMA (D4K9N, #19245, Cell Signaling Technology Inc. 1:200) or anti-lyve-1 (NB100–725B, NOVUS biologicals, 1:200) antibodies, and then processed with the ImmPRESS™ HRP Universal Antibody (Horse Anti-Mouse/Rabbit IgG) Polymer Detection Kit (MP-7500-50, Vector Laboratories, Inc.) and Liquid DAB+ Substrate Chromogen System kit (K3468, DAKO). Slides were counterstained with hematoxylin and dehydrated through graded alcohols and xylene. Mounting media were added to seal slides with coverslips (29).Images were captured by microscope (Nikon, Tokyo, Japan). CK7/Ki-67 positive in lymph node was diagnosed as metastasis (30). Data were analyzed by Fiji (ImageJ, NIH).

Following deparaffinization, rehydration, antigen retrieval, permeabilization (0.5% Triton X-100 in PBS) and non-specific binding blocking (2.5% normal horse serum, MP-7500-50, Vector Laboratories, Inc.; 2.5% normal goat serum, 50062z, Thermo Fisher; 2.5% bovine serum albumin solution, Sigma), slides were stained with anti-α-SMA (D4K9N, #19245, Cell Signaling Technology Inc.; 1A4, ab7817, abcam,1:200) and anti-lyve-1 (NB100–725B, NOVUS biologicals, 1:500) antibodies overnight at 4°C. Alexa Fluro™ 568 donkey anti-mouse IgG (H+L) (#A-10037, Invitrogen; Thermo Fisher Scientific, Inc.) and Alexa Fluro™ 488 goat anti-rabbit IgG (#A-10034, Invitrogen; Thermo Fisher Scientific, Inc.) were diluted in blocking buffer and incubated at 4°C in the dark for 1 hour. Prolong^® Gold anti-fade reagent (Invitrogen; Thermo Fisher Scientific, Inc.) containing DAPI was added to the cover slips. Slides were sealed with nail polish. Images were captured by a fluorescence microscope (Nikon, Tokyo, Japan). α-SMA scoring was evaluated as an “Area Index,” calculated by Fiji (ImageJ, NIH). To ensure that the whole tumor was evenly evaluated, at least three or more fields including stromal cells were carefully selected to evaluate CAFs. The mean value obtained from each sectioned tissue was defined as the α-SMA⁺ area index (31). The hot spots of lyve-1 positive lymph vessels were measured as lymphangiogenesis at 200x magnification (29).

Fibrillar collagen deposition was conducted using Picrosirius red staining kit (#24901, Polysciences, Inc.) (32). Images were captured by microscope (Nikon, Tokyo, Japan). Collagen deposition (red) was evaluated as an “Area Index,” calculated by Fiji (ImageJ, NIH).

Around 5000 cells were seeded in a 96-well plate and incubated in a humidified atmosphere of 5% CO2 at 37°C for 24 hours (26). ABT-199 was added at concentrations from 0-40 μM and incubated for 24 hours. 20 μl of 2.5 mg/ml MTT was added to each well and incubated at 37°C for 1 hour. Media were carefully removed and 100 μl DMSO added to each well. Absorbance was read at OD590 nm.

Tumor formation was confirmed by micro-CT 7 days after injection. Mice were and randomly assigned to two groups and treated with vehicle or ABT-199 (100 mg/kg) (33) daily by oral gavage for 2 weeks, 4 weeks or 8 weeks (33, 34). ABT-199 (Venetoclax) was purchased from MedChemExpress (MCE^® Cat. No.: HY-15531/CS-1155, Purity:99.96%). ABT-199 (10 mg/ml) was formulated in 5% DMSO, 40% PEG 300, 5% Tween 80, and 50% dH2O.

Data collected from triplicate experiments are presented as Mean ± SEM. Comparison of different groups was performed using Student’s paired or unpaired two-tailed t-test, one-way ANOVA and two-way ANOVA by GraphPad Prism (v6.0). Statistical significance (n.s.: not significant, *: p <0.05, **: p <0.01, ***: p <0.001, ****: p <0.0001) indicates comparison with control.

To investigate whether human lung fibroblasts can be transformed into CAFs by cancer cells, we co-cultured MRC-5 cells with H358, HCC827, and H1993 cells for 14 days. Following co-culture, MRC-5 cells exhibited significant morphological alterations, including prominent filament formation, as observed by light microscopy ( Supplementary Figure S1a ) and immunofluorescence (IF) microscopy ( Supplementary Figure S1b ), compared to monocultured MRC-5 controls ( Supplementary Figures S1a–c ).

IF analysis revealed elevated expression of α-SMA and collagen type 1A1 (Col 1A1) in MRC-5 cells co-cultured with H358 and HCC827 cells ( Figure 1a ), whereas PDGFRβ (21) and PDPN (8)—markers often associated with CAFs—were not detected ( Supplementary Figure S1e ). Flow cytometry confirmed an increase in α-SMA+ cells in the H358 + MRC-5 co-culture system, with ~20% of fibroblasts acquiring this CAFs marker (21) ( Figure 1b ). Notably, VEGF-C, a key mediator of tumor lymphangiogenesis (35), was markedly upregulated in α-SMA+ CAFs, while VEGF-D (36) remained undetectable ( Figure 1a , Supplementary Figure S1e ).

To further characterize these activated fibroblasts, we isolated α-SMA+ CAFs by cell sorting ( Figure 1c ), hereafter designated as H358-CAFs and HCC827-CAFs. Western blot analysis confirmed significant upregulation of α-SMA and Col 1A1 (37–39) in H358-CAFs and HCC827-CAFs compared to parental MRC-5 cells ( Figure 1d ), indicating not only phenotypic transformation but also functional activation, as evidenced by increased collagen production.

Collectively, these findings demonstrate that H358 and HCC827 cells effectively reprogram normal lung fibroblasts into CAFs within 14 days of direct co-culture ( Figure 1e ). The resulting H358-CAFs and HCC827-CAFs exhibit phenotypic and functional properties consistent with primary tumor-derived CAFs (9, 21, 40), supporting their potential use in CAFs-rich models for orthotopic xenograft studies.

To evaluate the functional role of H358-CAFs in vivo, we performed direct intrathoracic injections of H358 cells, with or without H358-CAFs, into the left lung of nude mice ( Figure 2a ). Prior to inoculation, H358-CAFs activation was confirmed by immunofluorescence (IF) ( Figure 2b ). Tumor formation and progression were monitored longitudinally using micro-CT imaging ( Supplementary Figure S2h ), validating this approach for tracking both initial tumor establishment and subsequent growth dynamics (27).

Histopathological analysis of resected left lung tumors ( Figure 2c ) revealed significantly increased fibrotic stroma in H358 + H358-CAFs tumors compared to H358-only controls ( Supplementary Figure S2i , Figure 2f ). Notably, H358-CAFs maintained persistence in the lung microenvironment throughout the experimental period, contrasting with PDX models where CAFs are typically lost (41). Immunohistochemical (IHC) and IF analyses demonstrated substantially higher α-SMA expression in H358 + H358-CAFs tumors (15-20%) versus H358-only (5%) or H358 + MRC-5 (5-10%) groups ( Figures 2d, e, h ), establishing the necessity of incorporating activated CAFs in orthotopic xenograft (OX) models (13). Similar α-SMA enhancement was observed in HCC827-CAFs co-injection experiments ( Supplementary Figure S2d ).

CAFs-mediated desmoplasia (20, 42), characterized by excessive fibrillar collagen deposition in the TME (38, 43), underscores their functional significance. Building on our in vitro findings of CAFs-mediated collagen production ( Figures 1a, b ), we assessed desmoplastic remodeling in vivo using picrosirius red staining (PRS). Tumors co-injected with H358-CAFs or HCC827-CAFs exhibited pronounced fibrillar collagen deposition compared to cancer cell-only controls ( Figure 2g , Supplementary Figure S2e ), demonstrating CAFs’ pivotal role in driving extracellular matrix (ECM) remodeling and desmoplasia in the TME.

Proliferation analysis revealed a significantly elevated Ki-67 index in H358 + H358-CAFs tumors (20-22%) versus H358-only controls (15-18%) ( Figure 2i ), indicating CAFs-mediated enhancement of cancer cell proliferation. Intriguingly, tumors with normal fibroblasts (H358 + MRC-5) showed reduced proliferation (14-16%), suggesting tumor-suppressive effects of untransformed fibroblasts (5). These observations underscore the preferential use of activated CAFs but not normal fibroblasts in enhancing the fidelity of the OX model.

Notably, IF analysis demonstrated substantial Lyve-1+ lymphatic vessel expansion in peritumoral regions of H358-CAFs co-injected tumors ( Figure 2e ), with parallel findings in the HCC827-CAFs model ( Supplementary Figure S2d ). Stromal α-SMA+ CAFs in both models showed strong VEGF-C upregulation ( Figure 2d , Supplementary Figure S2g ) but no detectable VEGF-D ( Supplementary Figure S2j ), implicating CAFs-derived VEGF-C as a key mediator of tumor-associated lymphangiogenesis. These observations further underscore the involvement of lung CAFs in promoting lymphangiogenesis (44).

Collectively, these results demonstrate that lung CAFs functionally remodel the tumor microenvironment through three synergistic mechanisms: (1) direct enhancement of cancer cell proliferation, (2) desmoplastic ECM remodeling via collagen deposition (45), and (3) VEGF-C-mediated lymphangiogenesis (7, 44). This triad of CAFs-driven effects establishes a pro-metastatic niche that facilitates mediastinal lymph node dissemination in NSCLC.

To evaluate the role of lung CAFs in facilitating lymphatic metastasis, we extended our analysis to 6 weeks based on previous reports of detectable lymph node metastases in H358 orthotopic xenografts at 4 weeks (16). Serial micro-CT imaging confirmed progressive tumor growth throughout the study period ( Figures 3a, b ).

Histopathological analysis revealed significantly increased α-SMA+ CAFs infiltration and Lyve-1+ lymphatic vessel density in primary tumors (pT) from H358-CAFs co-injected xenografts compared to H358-only controls ( Figure 3c ), confirming the establishment of robust CAFs-rich, lymphangiogenic tumor models. Most notably, both H358-CAFs and HCC827-CAFs co-injection groups demonstrated substantially higher incidence and burden of mediastinal lymph node metastases compared to respective cancer cell-only xenografts ( Figures 3d, e , Supplementary Figure S2f ).

Enhanced apoptotic sensitivity has been extensively documented in various activated cell states, such as α-SMA⁺ fibroblasts (46, 47). Upon differentiation from normal fibroblasts, activated α-SMA⁺ fibroblasts exhibit heightened mitochondrial priming, rendering them more susceptible to apoptosis and relying more on anti-apoptotic proteins for survival (46, 48). This heightened apoptotic sensitivity is closely linked to their enhanced ability to initiate the mitochondrial pathway of apoptosis, primarily governed by the Bcl-2 family of proteins (49, 50). In line with these findings, ABT-199, a Bcl-2 inhibitor known for its pro-apoptotic effects, has been developed and demonstrated efficacy in eliminating α-SMA⁺ fibroblasts (34, 51). Nevertheless, its effectiveness in depleting lung CAFs in vitro remains to be fully elucidated.

Our MTT assays revealed a striking differential sensitivity: while H358, HCC827, and MRC-5 cells showed minimal response, both H358-CAFs and HCC827-CAFs demonstrated dose-dependent apoptosis ( Figures 4a, b ). This selectivity was confirmed in co-culture systems, where 24-hour ABT-199 treatment eliminated most α-SMA+ CAFs while sparing cancer cells ( Figure 4c ). Flow cytometry quantified this effect, showing significant depletion of α-SMA+ populations ( Figure 4d ). Importantly, normal MRC-5 fibroblasts maintained proliferation rates despite treatment ( Figure 4e ), confirming the therapeutic window between activated CAFs and untransformed cells.

To investigate the in vivo impact of ABT-199 (100mg/kg) (33), we conducted experiments using the H358 + H358-CAFs OX model ( Figure 5a ). Analysis via micro-CT revealed a significant reduction in tumor size after a 14-day treatment with ABT-199 ( Figure 5b ), indicating its potential to suppress tumor growth. Additionally, IHC analysis demonstrated a marked decrease in Ki-67 index in orthotopic xenografts following ABT-199 administration ( Figure 5c ), suggesting decreased proliferation upon elimination of lung CAFs. Histological examination also illustrated a notable decline in both α-SMA and lyve-1 expression levels in ABT-199-treated H358 + H358-CAFs orthotopic xenografts compared to controls ( Figure 5d ). Additionally, IHC analysis revealed diminished VEGF-C expression in the stromal area of orthotopic xenografts ( Figure 5e ), indicating reduced lymphangiogenic growth factors within the TME following depletion of lung CAFs. These coordinated effects demonstrate that ABT-199-mediated CAFs elimination disrupts the pro-tumorigenic microenvironment, impairing both tumor growth and lymphatic metastasis potential.

The observation of only 1–2 mediastinal lymph node metastases in the 6-week H358-only OX model ( Figure 3e ) raised concerns regarding the limited robustness for comparative analysis or definitive conclusions. To address this, we extended the experimental duration to 9 weeks in both the H358 + H358-CAFs and H358-only OX models ( Figure 6a ) to increase the incidence of lymph node metastasis in the control group. Additionally, we initiated ABT-199 treatment at week 1 post-tumor formation and sustained it for 8 weeks (33) in both OX models to assess its potential in inhibiting mediastinal lymph node metastasis ( Figure 6a ).

Longitudinal micro-CT analysis revealed a significant reduction in tumor diameters from weeks 5–8 in the H358 + H358-CAFs group, while no such effect was observed in H358-only xenografts ( Figure 6b ). Importantly, Ki-67 indices remained unchanged in both models ( Figure 6c ), confirming that ABT-199’s antitumor effects were mediated through microenvironmental modulation rather than direct cancer cell cytotoxicity.

PRS analysis demonstrated progressive stromal activation, with fibrillar collagen deposition increasing from <5% at 2 weeks to 5-10% by 9 weeks in H358-only tumors ( Figure 6d ). The H358 + CAFs group showed significantly greater collagen accumulation, highlighting the sustained pro-fibrotic activity of transplanted CAFs. ABT-199 treatment substantially reduced collagen deposition in both models and significantly decreased α-SMA+ areas in H358 + CAFs xenografts ( Figures 6d, e ), demonstrating effective CAFs depletion and consequent disruption of desmoplastic stroma.

Metastatic analysis revealed ABT-199’s potent anti-metastatic activity, with significantly reduced mediastinal lymph node involvement in both models ( Figures 6f-h ). The treatment maintained an excellent safety profile, with no significant body weight changes or evidence of liver toxicity (33) ( Figures 6i, j ).

These findings establish that ABT-199 inhibits metastasis primarily through CAFs elimination and subsequent collapse of the pro-metastatic niche, rather than through direct tumor cell effects ( Figure 7 ). The compound’s specific activity against CAFs-rich tumors and favorable toxicity profile position it as a promising stroma-targeted therapeutic for preventing NSCLC lymphatic dissemination.