All samples were collected at the First Affiliated Hospital of Nanchang University and all procedures were approved by the ethics committee of the hospital (approval no. 017). Informed patient consent was obtained before surgery. Nine HNSCCs were surgically removed, divided into 2 × 2 × 2 mm³ sections, and placed in 96-well plates (one block/well) at a low temperature. After injection of a water-containing gel, the tissues were cultured at 37°C at 95% humidity under 5% (v/v) CO₂. Three-quarters of the medium containing different drugs was changed every 1–2 days. The control cells were treated with Dulbecco’s modified Eagle medium/F12 complete medium (Cat. No.11330107; Gibco, Amarillo, TX, United States) supplemented with 10% (v/v) fetal bovine serum (Cat. No.16140063; Gibco), 100 units/ml penicillin, and 100 μg/ml streptomycin. The tissues in the control groups were cultured in a manner similar to control cells. The drug groups were cultured in complete medium with 3% (v/v) dimethyl sulfoxide (DMSO, Cat. No. D4540; Sigma, St. Louis, MO, United States), 24 µM anlotinib (Zhengda Tianqing Pharmaceutical Group Co. Ltd., Nanjing, China), 20 µM sunitinib (Pfizer Pharmaceuticals Ltd., New York, NY, United States), 9 µM apatinib (Hengrui Pharmaceuticals, Jiangsu, China), 75 μg/ml paclitaxel (Beijing SL Pharmaceutical Co. Ltd., Beijing, China), or 25 μg/L cisplatin (Harvey, United States). The culture medium was replaced after 24 h to maintain the activity and morphology of the original tissue block. On day 8, the medium was removed, tissues were washed three times with phosphate-buffered saline, and MTT solution (Sigma) was added. After incubation for 16 h, the tissue blocks were placed in new 96-well plates, and each block was incubated with 200 µl DMSO for 6 h. Aliquots of 100 µl DMSO were then placed into the wells of a fresh 96-well plate, and the absorbance (A values) at 490 nm were measured. Drug sensitivity was assessed based on the tumor inhibition rate, which was calculated as [1—(drug A-blank A/control A-blank A)] × 100%; a tumor inhibition rate ≥30% was presumed to indicate drug sensitivity. The experiment was repeated three times and averages were calculated.

PDXs (numbered as P0) were established. Three PDX models, one of hypopharyngeal carcinoma and two of laryngeal carcinomas, were selected and passaged to P3 by the same method, then used for in vivo experiments. When the diameters of the third-generation subcutaneous tumors reached 1 cm, they were minced (2 × 2 × 2 mm³) and subcutaneously implanted into the flanks of 4- to 6-week-old nude mice. Each PDX was inoculated into 24 mice. When the tumor volumes reached 100–200 mm³, the mice were randomly divided into three groups of six: a control group (normal saline 100 μl intragastrically per day), a drug group (anlotinib 3 mg/kg intragastrically per day), and a positive control group (cisplatin 5 mg/kg intraperitoneally per week). Mouse health and tumor growth were observed daily. The tumor volume and mouse body weight were recorded every 4 days. Twenty-one days later, when the tumor volume in the control group reached 1,500 mm³, all mice were sacrificed via cervical dislocation, and the tumor tissues were harvested. Tumor size was measured using a caliper, and tumor volume was calculated as 1/2 × length × (width)². Tumor growth inhibition (TGI) was used to evaluate the anti-tumor effects of anlotinib. TGI (%) was calculated as: TGI (%) = [1—(tumor volume at the end of the drug treatment group—tumor volume at the beginning of the drug treatment group)/(tumor volume at the end of the control group—tumor volume at the beginning of the control group)] × 100%. All experiments were approved by the Animal Ethics Committee of Nanchang Leyou Biotechnology Co. Ltd.

PDX construction was performed using the protocol established by Karamboulas et al. (19). Briefly, HNSCC-PDXs were generated, the tumor tissues were processed, and the PDX models were expanded into cohorts for drug testing.

Formalin-fixed PDX samples were embedded in paraffin and cut into 3-µm-thick sections that were baked at 60°C for 12 h. The sections were then placed into xylene for 5–10 min (repeated three times), hydrated in 100, 100, 95, 85, and 75% (w/w) ethanol (in sequence) for 5 min each, soaked in distilled water for 5 min, stained with hematoxylin (Wellbio, China) for 5–10 min, washed with distilled water, stained with eosin (Wellbio) for 3–5 min, washed with distilled water, dehydrated in 95 and 100% (v/v) alcohol baths for 5 min each, placed in xylene for 10 min (two repeats), mounted with neutral gum, and observed microscopically.

After dewaxing, some tissues were used for immunohistochemistry and TUNEL assays. Some slides were subjected to Proteinase K incubation and used for permeabilization; positive slides were incubated with recombinant terminal deoxynucleotidyl transferase enzyme, Equilibration Buffer, dUTP Labeling Mix, whereas negative slides were incubated with distilled water instead of recombinant terminal deoxynucleotidyl transferase enzyme. The TUNEL Apoptosis Detection Kit featuring fluorescein isothiocyanate (Cat. No. 40306ES20) was purchased from Shanghai YEASEN (Shanghai, China). DAPI (Cat. No. 28718-90-3) was obtained from Proteintech (Rosemont, IL, United States). Apoptosis was evaluated after drug treatment in vivo, following the manufacturer’s instructions.

The TUNEL assay was performed as described by Loo (20). Briefly, tissues were fixed in formaldehyde and permeabilized with Proteinase K to allow the penetration of TUNEL reaction reagents into the cell nucleus. After fixation and washing, the incorporation of biotinylated dUTP onto the 3′ ends of fragmented DNA was conducted in a reaction containing terminal deoxynucleotidyl transferase. The slides were mounted with 90% glycerol and visualized by fluorescence microscopy. The TUNEL stained slides were calculated by manual counting under a fluorescent microscope. The TUNEL-stained cells appeared green, while the other cell nuclei appeared blue. The apoptosis index (AI) was regarded as the number of TUNEL-positive cells/total number of cells × 100%.

Immunohistochemistry (IHC) was performed as previously described (21). Briefly, paraffin-embedded xenografts or tissues were cut into 3-μm sections and mounted on slides, subjected to antigen retrieval in hot citrate buffer (pH = 7.4), then blocked in 10% normal goat serum. Subsequently, sections were incubated with primary antibodies against Ki-67 or proliferating cell nuclear antigen (Cat. Nos. 27309-1-AP and 10205-2-AP PCNA; Proteintech). The slides were incubated with goat anti-mouse/rabbit poly-horseradish peroxidase-conjugated secondary antibodies (Cat. No. PR30009; Proteintech). Nuclear Ki-67, as well as nuclear and cytoplasmic PCNA, stained red clay or brown. The PCNA slides were scored and analyzed using Image-Pro Plus software. Six 400X images were analyzed for each slide. The average integrated optical density was calculated as the ratio of the positive area to the total area. The Ki67 slides were analyzed using a microscope. Six 400X images were analyzed for each slide. The positivity cell rate was calculated as the ratio of the number of positive cells to the number of total cells.

Statistical analyses were performed using the SPSS software package (ver. 22.0; IBM, United States). All graphs were produced using Graph Pad Prism ver. 7.0 for Windows (Graph Pad Software Inc., United States). All experiments were performed in triplicate, with the means of those three independent experiments calculated. Data are presented as the means ± standard deviations (SDs) and were compared using Student’s t-test (two groups) or one-way ANOVA (multiple groups). A p-value < 0.05 was considered to indicate statistical significance.

Tumor samples from nine HNSCC patients were cultured in vitro at 37°C under 5% (v/v) CO₂ (Table 1). The samples were labeled as Control, DMSO, anlotinib, sunitinib, apatinib, paclitaxel, and cisplatin. We used the MTT assay to assess tumor cell viability; anlotinib significantly inhibited cell viability (Figure 1A; Table 1). For all nine samples, anlotinib exhibited a greater inhibitory effect than did sunitinib and apatinib. The tumor inhibition rates were as follows: DMSO, 10.87 ± 6.58%; paclitaxel, 50.05 ± 11.33%; cisplatin, 41.88 ± 8.55%; anlotinib, 51.05 ± 13.74%; sunitinib, 30.03 ± 12.85%; and apatinib, 25.06 ± 16.37% (Figure 1B). Because drug inhibition rate >30% was regarded as an indicator of drug sensitivity, anlotinib treatment presumably resulted in significant tumor suppression in vitro.

Given the in vitro data, nine PDXs (numbered as P0) were established. Three PDX models were selected and passaged to P3 by the same method, then used for in vivo experiments. Mice bearing the PDXs were treated with drugs for 21 days, at which time the tumor volume in the control groups reached 1,500 mm³. The mice were then sacrificed. No mouse lost more than 10% of its body weight during the experiment, and no abnormal behavior was noted. Anlotinib was significantly more inhibitory (TGI 79.02%) than cisplatin (TGI 40.57%) (Figure 2A). The tumor volume in the anlotinib group on day 21 did not differ significantly from the initial volume (Figure 2B). The tumors in the anlotinib group were smooth-surfaced (unlike the tumors in the other two groups).

To evaluate the anti-tumor effects of anlotinib, the Ki-67 and PCNA indices of tumor sections were immunohistochemically evaluated. The Ki-67 and PCNA expression levels differed significantly among the three groups (Figure 3A). In the anlotinib group, the Ki-67 positive rates were 20.26 ± 1.60% for patient no. 8, 26.7 ± 4.47% for patient no. 10, and 25.13 ± 3.24% for patient no. 17; these were significantly lower than the rates in the control group (61.89 ± 4.44% for patient no. 08, 55.16 ± 6.17% for patient no. 10, and 55.85 ± 2.23% for patient no.17; p < 0.05) (Figure 3B). Similarly, compared to the control group, the PCNA-average integrated optical densities of 0.19 ± 0.01, 0.15 ± 0.01, and 0.10 ± 0.01 for patients no. 8, 10, and 17, respectively, were significantly lower than (p < 0.05) the corresponding figures in the control group (0.50 ± 0.01, 0.35 ± 0.01, and 0.21 ± 0.01) (Figure 3C). The Ki-67-positivity rates and PCNA-average integrated optical densities were lower in the anlotinib group than in the positive control group, suggesting that tumor growth was inhibited in HNSCC PDX mice; moreover, anlotinib significantly inhibited tumor cell proliferation.

Hematoxylin-and-eosin staining of the HNSCC tumor tissues of three tumor-bearing mice showed that the extents of necrosis and apoptosis varied among the groups. Tumors actively proliferated in the control group, but less so in the positive control and anlotinib groups. Apoptosis was maximal in the anlotinib group, in which karyopyknosis, karyorrhexis, and apoptotic bodies were observed (Figure 4A). TUNEL assays were performed, with green fluorescence indicating apoptosis and nuclei fluorescing blue. The highest apoptosis levels were observed in the anlotinib group (Figure 4B) (patient no. 8: 27.03 ± 1.09%, no. 10: 28.69 ± 0.96%, and no. 17: 41.12 ± 1.21%; the control values were no. 8: 5.01 ± 0.63%, no. 10: 5.245 ± 0.33%, and no. 17: 9.69 ± 0.52%; and the positive control values were no. 8: 9.50 ± 0.58%, no. 10: 11.50 ± 0.71%, and no. 17: 9.94 ± 1.648%) (p < 0.05) (Figure 4C). Therefore, anlotinib promoted tumor cell apoptosis.