Therapeutic Evaluation of Palbociclib and Its Compatibility with Other Chemotherapies in Primary and Recurrent Nasopharyngeal Carcinoma

Background: Recent genomic analyses revealed that druggable molecule targets could only be detected in around 6% of nasopharyngeal carcinoma (NPC) patients. Yet, an addiction to dysregulated CDK4/6-cyclinD1 signalling pathway is an essential event in the pathogenesis of NPC. Using our newly established xenografts and cell lines derived from primary, recurrent and metastatic NPC, we aimed to evaluate the therapeutic ecacy of a specic CDK4/6 inhibitor, palbociclib, and its compatibility with other chemodrugs in treating NPC. Methods: The ecacy of single treatment of palbociclib on NPC models was rst evaluated, followed by concurrent treatment with cisplatin or suberanilohydroxamic acid (SAHA). RNA sequencing was used to prole the related pathways in governing the drug response. Palbociclib-resistant NPC cell lines were also established to demonstrate if cisplatin could be used as a second-line treatment once the cells developed resistance to palbociclib. The ecacy of palbociclib treatment on cisplatin-resistant NPC cells was also examined. Results: Palbociclib single drug treatment was conrmed to have a cell cycle arresting effect of NPC cells in G1 phase in vitro. It also had a signicant inhibitory effect in all the 6 NPC tumor models in vivo, with a substantial reduction in total tumor volume and proliferation marker Ki-67. Concurrent use of palbociclib dampened the cytotoxic effect of cisplatin in NPC cells in vitro. Notably, combination of palbociclib with SAHA resulted in synergistic cell death of NPC both in vitro and in vivo. Autophagy-associated cell death was found to be involved in the enhanced tumor growth inhibitory effect in the combined palbociclib+SAHA treatment. NPC cell lines trained to sustain growth in high dose of palbociclib and cisplatin remained sensitive in subsequent treatment of cisplatin or palbociclib respectively. Conclusions: This study provides essential evidences to position palbociclib as an alternative therapeutic option to NPC treatment, and to aware the effective administrative timing of palbociclib with other chemodrugs. The ndings give the basis for planning of the rst-in-human clinical trials of palbociclib regimens in NPC patients. cyclin D1 and downregulation of p16 are common, supporting the druggability of NPC by CDK4/6 specic inhibitor targeting this essential cell cycle regulatory pathway Our earlier study has shown that overexpression of cyclin D1 is present in over 90% of tumor tissues (18). Cyclin D1 is also overexpressed in premalignant and dysplastic nasopharyngeal epithelial cells (NPE) and may play an important role in early pathogenesis of NPC by supporting persistent and latent EBV infection in premalignant nasopharyngeal epithelium (19). Whole exome sequencing analysis of a large cohort of NPC (n > 100) has also identied that amplication of cyclin D1 and homozygous deletion of p16 gene are common features of NPC (10). Importantly, RB mutation is uncommon in NPC, suggesting the blockage of CDK4/6 activity can abolish the dependency of NPC in the cyclin D1/RB pathway for tumor growth. newly established EBV positive NPC cell lines by our laboratory. carefully characterized with respect to EBV infection, genomic proles and growth properties. The NPC cell lines maintained as monolayer in RPMI with 10% FBS and supplemented with 1% penicillin/streptomycin and examined for their responses to palbociclib in both 2-D (monolayer) and 3-D (spheroid) conditions. For 3-D culture, the NPC cells were seeded in ultra-low attachment plates (Corning, #4520) to form oating 3D spheroids. µM palbociclib, A cyclin E1, cyclin D2, and cyclin D3 in the resistant NPC43 cells. All these results indicate the resistant cells may have acquired alternate pathways, which are independent of CDK4/6/cyclin D1/RB pathway, to maintain cell proliferation under the presence of palbociclib. Moreover, the resistant lines also had a downregulated E-cadherin expression, but an upregulated N-cadherin expression, suggesting the resistant line may be more prone to metastasis. We also the expression of the cancer stemness related genes. MMP2, MMP9, Nanog, and SOX2 mRNA were identied to be elevated through the qPCR test These results suggested that palbociclib-resistant NPC43 cells may acquire an increase in cancer stemness. an clinical use, effect of signicant trial, we did not observe signicant difference in body weights between mice in the and control groups. we could inhibit the lung colonization of NPC cells in vivo after tail-vein injection suggesting its potential inhibitory effect in NPC metastasis S5). In breast cancer, palbociclib was shown to inhibit metastasis in animal models through inhibition of the c-Jun/COX-2 signaling pathway (48). The detail molecular mechanism of how palbociclib suppresses metastasis of NPC warrants more investigation. The potent suppressive effect of palbociclib on growth of NPC xenografts derived from NPC patients with primary, recurrent and metastatic cancer supports its application in the clinical trials of NPC treatment. A previous preclinical study using another CDK4/6 inhibitor, ribociclib, which interferes a different site in the ATP-binding pocket of CDK4/6 as that of palbociclib, also demonstrated the inhibition of CDK4/6 could inhibit the growth of NPC cells (49). An integrated genomic and transcriptomic study of 5 patient-derived xenografts also discovered the copy number of CCND1 and CDKN2A are the potential drug target of palbociclib to suppress tumor growth (50). There was also a case report demonstrating the clinical benets from palbociclib in a patient with previously treated metastatic NPC with CDK4 amplication (51). All these observations indicate that targeting the dependency of NPC cells in this cell cycle signaling pathway is of high therapeutic value to NPC patients. the A combined regimen of the inhibitory effect of palbociclib. Concurrent use of cisplatin and palbociclib is not advised. NPC cells which developed tolerance of palbociclib remain sensitive to cisplatin, and vice versa. Together, the work herein provides relevant information for planning clinical application of palbociclib-involved regimens in NPC treatment.


Background
NPC is prevalent in southern China and Southeast Asia. Platinum-based chemotherapy in combination with radiotherapy is the mainstay treatment of primary and local NPC (1). However, local recurrence still occurs in 5 to 10% of NPC patients after treatment, and in 15 to 45% Stage IV patients. Around a quarter of Stage IV NPC patients may also develop distant metastasis (2,3). Management of advanced NPC, including recurrence, metastasis and chemo-resistant tumors, remains a major challenge. Therefore, developing a new modality of targeted therapy effective to control NPC at both early and advanced stages is eminent and highly desirable to improve the survival of NPC patients.
The lack of preclinical NPC model is one of the major obstacles limiting screening and evaluation of novel and potentially effective therapeutic agents against NPC. NPC patient-derived xenografts and cell lines are known to be di cult to establish. Hence, very few NPC xenografts were available for investigation in the past which include the XenoC15 and XenoC17 (derived from African NPC patients; (4)), Xeno2117 and Xeno666 (derived from HK NPC patient; (5)). However, these NPC xenografts were established over 25 years ago and have been passaged for long period of time (over 25 years) in nude mice. Their genetic and pathological properties may have drifted away from their parental NPC tumours in patients. For in vitro cell lines, the situation is even worse. There was only one EBV-positive NPC cell line (C666-1) available for investigation (6). The other commonly used "NPC cells" have all lost their Epstein-Barr virus (EBV) episomes and may not be representative of NPC (7). Furthermore, the presence of genetic component of HeLa cells as well as the presence of HPV18 genome in these cell lines also casted ambiguity to their cellular origins (7) which limited their applications in evaluation of novel therapeutic agents against NPC. To relieve the limitation of current preclinical models of NPC (both in vivo and in vitro) for investigation, we have established new NPC xenografts and cell lines, including Xeno32 and Xeno76 (xenografts derived from primary NPC (8)); Xeno23 and NPC43 (xenograft and cell line respectively derived from recurrent NPC (8)); C17 (NPC cell line derived from xenograft of a metastatic NPC (9)). These newly established NPC xenografts and cell lines, in conjunction with the conventional NPC cell line C666-1, represent a comprehensive panel of preclinical NPC models available for assessing the drug e cacy against NPC.
Targeted therapy is grossly underdeveloped in NPC. Apart from the scarcity of representative preclinical NPC models for evaluation of novel therapeutically agents, it is also due to our insu cient knowledge on the genomic properties of NPC. Recently, we and others have conducted several genomic analyses to de ne the genetic alterations contributive to NPC tumorigenesis (10)(11)(12). All these knowledges will shape the focus of future development of NPC therapy (13). At present, genetic alterations amenable to targeted therapy are heterogenous and of relative low mutation rates in NPC. Mutations in PIK3CA, EGFR, FGFR1/2/3/4, BRCA1/BRCA2/ATM were only indenti ed in around 1.68%, 0.24%, 2.16%, 1.68% respectively in NPC patients (10,11,14,15). There are evidences indicating that NPC cells are addicted to dysregulated p16-CDK4/6-cyclin D1-RB signaling pathway for their aggressive growth and metastatic behaviors. In proliferative cells, expression of p16 is commonly suppressed to relieve its inhibitory effect on the kinase activity of CDK4/6. The CDK4/6 kinases form an active complex with cyclin D which hyper-phosphorylates RB and releases the E2F leading to a cascade of downstream events involved in the transcription of proliferation genes, enabling cells to enter cell cycle. In NPC, overexpression of cyclin D1 and downregulation of p16 are common, supporting the druggability of NPC by CDK4/6 speci c inhibitor targeting this essential cell cycle regulatory pathway (16)(17)(18). Our earlier study has shown that overexpression of cyclin D1 is present in over 90% of tumor tissues (18). Cyclin D1 is also overexpressed in premalignant and dysplastic nasopharyngeal epithelial cells (NPE) and may play an important role in early pathogenesis of NPC by supporting persistent and latent EBV infection in premalignant nasopharyngeal epithelium (19). Whole exome sequencing analysis of a large cohort of NPC (n > 100) has also identi ed that ampli cation of cyclin D1 and homozygous deletion of p16 gene are common features of NPC (10). Importantly, RB mutation is uncommon in NPC, suggesting the blockage of CDK4/6 activity can abolish the dependency of NPC in the cyclin D1/RB pathway for tumor growth.
In this study, we have used the comprehensive panel of NPC models to examine the e cacy of FDA-approved palbociclib, a selective CDK4/6 inhibitor, in the treatment for NPC. Palbociclib has been shown in multiple studies to be effective against cancer cells which have (a) overexpression of cyclin D1, (b) functional RB and (c) inactivated p16 (20)(21)(22). As described before, this signaling axis of CDK4/6-cyclinD1 is commonly dysregulated in early and advance stages of NPC. Inhibition to the CDK4/6 activity represents a common target for NPC patients with primary, recurrent and even metastastic tumors. In this study, we have also examined the treatment e ciency of combination treatment of palbociclib with other FDA-approved chemotherapies, cisplatin and suberanilohydroxamic acid (SAHA). Interestingly, we observed a synergistic effect in combination treatment of SAHA with palbociblib. SAHA is a histone deacetylase (HDAC) inhibitor which alters gene transcription through inhibition of deacetylation of histones and induces chromatin relaxation leading to general expression of tumor suppression genes (23). The synergistic effect of combined treatment of SAHA and palbociclib was also con rmed for the rst time in our preclinical models of NPC xenografts. Transcriptome pro ling of the NPC cells treated by SAHA and palbociclib revealed the activation of autophagy may be involved in their synergistic actions to inhibit the growth of NPC cells. Furthermore, we have established palbociclib-resistance and cisplatin-resistance cell lines to evaluate their drug response to cisplatin and palbociclib respectively. These ndings provide essential information for planning of the rst in-human trial of palbociclib regimens in the treatment of NPC.

Materials & Methods
Non-malignant NPE and cancerous NPC cell lines Three telomerase-immortalized nonmalignant human NPE cell lines (NP361hTert, NP460hTert and NP550) (24,25), and one SV40T-immortalized NPE cell line (NP69) (26) were used in this study as non-cancer cell controls. They were cultured in conditions described in our previous publication (27). Three EBV-positive NPC cell lines (C666-1, NPC43, and C17) were used in this study. C666-1 (6) was established from an NPC xenograft, XenoC666, which is widely used in NPC preclinical studies. NPC43 (8) and C17 (9) are newly established EBV positive NPC cell lines by our laboratory. They have been carefully characterized with respect to EBV infection, genomic pro les and growth properties. The NPC cell lines were maintained as monolayer in RPMI with 10% FBS and supplemented with 1% penicillin/streptomycin and examined for their responses to palbociclib in both 2-D (monolayer) and 3-D (spheroid) conditions. For 3-D culture, the NPC cells were seeded in ultra-low attachment plates (Corning, #4520) to form oating 3D spheroids.

NPC xenografts
Four-week-old male immune-de cient mice (NOD/SCID) were supplied by the Laboratory Animal Unit (LAU) of The Hong Kong University (HKU) and housed under pathogen-free conditions. All animal experiments were conducted in conditions according to the animal license issued from the Hong Kong Department of Health and with the approval of the Committee on the Use of Live Animals in Teaching and Research (CULATR) of HKU. To initiate the growth of NPC cell lines (C17, NPC43, and C666-1 cells) as tumor xenografts in NOD/SCID mice, we resuspended 1 × 10 7 cells in 200 µl of 1:1 mixture of Matrigel and culture medium and injected subcutaneously at the left side of the dorsal ank region of each mouse. For the three newly established xenografts, Xeno23, Xeno32, and Xeno76 (8), the xenografted tumors were cut into blocks of 2 mm size and implanted subcutaneously into the left side of the dorsal ank region of the mice. Mice were then randomized into drug treatment groups or vehicle control groups once the tumors become palpable (with a diameter of 4 mm). The CDK4/6 inhibitor, palbociclib (P zer, 571190-30-2) was dissolved in ltered ddH 2 O (7.5 mg/ml) and given to animals by daily oral gavage at the various concentrations stated in each experiment. SAHA (Cayman, 10009929) was dissolved in DMSO to a concentration of 100 mg/ml and then given to mice by intraperitoneal injection of 20 ul. Tumor size and animal body weight were recorded every other day during the entire treatment period. All the mice were euthanized at the end of the experiment. For histopathological examination and immunohistochemistry study, the tumors were excised and xed with 10% neutral buffered formalin (NBF).
Cell viability determination NPC cells were seeded at 4000 cells/well in a 96-well plate and incubated in 100 µl culture medium overnight. Palbociclib (Selleck Chemicals, S1116) was dissolved in cell culture medium in doses ranging from 0-20 µM and added to NPC cells. The viability of the cells was examined on day1, day3, and day5 respectively. SAHA was diluted from the stock solution (100 mg/ml) with the culture medium and used at various treatment concentrations ranging from 0 to 0.5 µM. Less than 0.001% of DMSO was used as a vehicle in the highest concentration of SAHA used in treatment. Cispaltin (Sigma, 479306) was diluted with dimethylformamide (DMF) at stock concentration 40 mM. Resazurin (Sigma, R7017) stock solution (0.02% w/v dissolved in PBS) was added to the cells (10% v/v) to determine the cell viability after drug treatment. After incubation the cells with resazurin for 4 hours, the uorescent product, resoru n, was measured at an excitation/emission of 530/590 nm using the Victor 3 Plate Reader (PerkinElmer). The growth inhibition was calculated as (viability control -vialbility drug )/ viability control *100%.
Western blot and qPCR analysis RIPA lysis buffer [50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% Nonidet P40, 0.5% deoxycholic acid, 0.1% sodium dodecyl sulfate (SDS)] was used to lyse the cells for Western blotting analysis. Protease inhibitors (1 tablet for 10 ml RIPA buffer) (Thermo Scienti c, A32961) and PhosSTOP (1 tablet for 10 ml RIPA buffer) (Roche, 4906837001) were included in the RIPA buffer immediately before use. Lysed protein samples were adjusted to equal protein concentration and resolved by SDS-polyacrylamide gel electrophoresis. PVDF membranes (GE Healthcare, 10500023) were used to transfer the resolved proteins in the gel.

Immunohistochemistry
Fresh tumor samples collected from NOD/SCID mice were xed in 10% neutral buffered formalin and processed for immunohistochemical examination.
Sections of 5-µm thick were cut from para n-embedded tumors and baked in 37ºC oven overnight before processing for H&E and immunohistochemical staining. The following speci c antibodies were used for immunohistochemical analyses: AE1/AE3 for keratin detection (1:250; DAKO, Cat. #: M3515) and Ki-67 for cell proliferation marker (1:200; Santa Cruz, Cat. #: SC-23900). For immunohistochemical staining, antigen retrieval was conducted. Slides were brought to boil in sodium citrate buffer (10 mM, pH 6.0) and maintained at just below boiling temperature for 20 minutes. Afterward, slides were de-waxed and rehydrated for para n sections. We rst incubated sections with 3% H 2 O 2 for 8 minutes to inactivate endogenous peroxidase. We then incubated sections with 3% bovine serum albumin for another 8 minutes to block the non-speci c binding. After overnight incubation with the primary antibody in a moist chamber, the horseradish peroxidase-conjugated secondary antibody (Dako, Cat. #: K4001) was applied to the sections for 1 hour at room temperature. The DAB (3,3'-diamino-benzidine; Dako, Cat. #: K346711-2) substrate was then applied to the sections for brown color development. The slides were then dehydrated and mounted with Permount mounting medium (Dako, Cat. #: S3023). The slides were then scanned and analyzed by the PE Vectra Polaris Imaging System.

RNAscope analysis to detect EBV gene expression in NPC xenografts
RNA in-situ hybridization protocol was conducted to examine the expression of selected EBV gene BZLF1(ACD, #450411) by using the RNAscope 2.5HD detection kit (Advanced Cell Diagnostic, # 322370) and following the recommended protocol from the manufacturer. This new RNA-in situ hybridization platform enables us to detect the expression of speci c EBV-RNA during lytic reactivation using speci c and sensitive RNA-FISH probe provided by the company.

RNAscope analysis to detect EBV gene expression in NPC xenografts
For examination of the RNA pro les of the NPC cells after drug treatments, mRNA libraries were prepared with TruSeq mRNA Library Prep kit (Illumina) and sequenced through the Illumina HiSeq2000 sequencing system (Illumina). The gene expression ratio between each treatment group was calculated based on the fragments per kilobase million values of each gene. HISAT2 was used to perform sequence alignment analysis using the reference sequence (28). Based on the alignment results of HISAT2, Stringtie (29) was used to complete the quantitative expression analysis of genes. Differential expression analysis was conducted by the edgeR (30). ClusterPro ler was used for KEGG and GO enrichment analysis.

Micro positron emission tomography (PET)/magnetic resonance imaging (MRI) scan
The mice were scanned using the nanoScan PET/3T MRI scanner (Mediso, Hungary) with 700 µm PET and 100 µm MRI spatial resolution. Prior to the scan, the mice were fasted overnight and each mouse received 200 µCi of 18F-urodeoxyglucose (FDG) through the tail vein with < 2 min of iso urane inhalation (5% in 100% oxygen). MRI scan and PET scan were performed 60 minutes after injection of 18F-FDG. The mice were placed into the PET/MRI scanner with a head holder under iso urane inhalation (2% in 100% oxygen) until the end of the scan. During the scan, their body temperature and respiratory rate were monitored. FDG uptake was quanti ed using standardized uptake values (SUVs). The SUV formula is as follows: SUV ¼ regional FDG concentration (Bq/mL)/injected FDG dose (Bq)* body weight (kg). The raw images were anatomically standardized to achieve symmetrical midline alignment. The images were reconstructed using the Nucline software (Mediso, Hungary) and PET/MRI fused images were coregistered using InterView FUSION (Mediso, Hungary). The internal liver metabolism (SUV liver = 0.5) was used as the basal metabolism level. The SUV of the tumor was normalized by basal metabolism. SUV = SUV tumor -SUV liver

Statistical analysis
Statistically analyzed results are presented as mean values ± SEM. Comparisons between groups were carried out by using a student T-test assuming equal variance and two-tailed distribution. Statistical analyses were performed using GraphPad Prism. p-value < 0.05 was considered as statistically signi cant.

Results
Palbociclib inhibited cell cycle progression in NPC cell lines Inactivation of p16 function, overexpression of cyclin D1, together with functional RB are established prediction markers for sensitivity of cancer cells to palbociclib (20)(21)(22), a speci c inhibitor targeting CDK4/6 (31). We rst examined the expression levels of p16, cylcin D1, phosphorylated RB protein and other relevant proteins in NPC cell lines and immortalized NPE cell lines to evaluate their potential sensitivity to palbociclib (Fig. 1A). The expression of these protein was examined in these cell lines grown in both 2-dimensional (2D) monolayer cultures and 3-dimensional (3D) spheroid cultures. Cell lysates were prepared from the culturing cells after plating in culture plates and grown for three days. The expression of p16 protein was only detected in the immortalized NPE cell line, NP69, but not in all the NPC cell lines (C666-1, C17 and NPC43). Functional RB status was examined by the presence of phosphorylation of RB at serine 780 residue (pRB-Ser780). The presence of pRB-Ser780 was detected in all the NPE and NPC cell lines, albeit at variable levels, but closely associated with the expression of cyclin A which can re ect the proliferation status of cells. Cyclin D1 and CDK4 were universally detected in all the NPE and NPC cells (Fig. 1A).
We then examined if there were differential sensitivities of NPC cell lines and non-malignant NPE cells grown in 2D conditions to palbociclib. Both NPC and NPE cell lines were treated with 0.2 µM of palbociclib for 24 hours and the changes in expression of pRB, total RB and cyclin A were monitored before and after treatment. As shown in Fig. 1B, treatment with palbociclib potently suppressed the levels of pRB in all the three NPC cell lines but not in immortalized NPE cells. Furthermore, expression of cyclin A, which is indicative of cells entering cell cycle, was universally suppressed in all three NPC cells after treatment with palbociclib. In contrast, expression of cyclin A was insigni cantly affected among the three immortalized NPE cell lines after treatment. Cell cycle analysis by ow cytometer further veri ed that 24-hour treatment with 0.2 µM of palbociclib induced signi cant G1 arrest in all the three NPC cell lines but had no signi cant impact on cell cycle progression in the three immortalized NPE cell lines (Fig. 1C). The % of G1-cells of the immortalized NP69, NP361 and NP460 only increased by around 0%, 14% and 6% respectively after palbociclib treatment; while that of C666-1, C17 and NPC43 drastically increased by 81%, 55% and 53% respectively after treatment. These results suggest that NPC cells are more susceptible to the inhibitory action of palbociclib than immortalized NPE cells.  Fig. S2). Suppression of cyclin A and pRB-Ser780 was also con rmed in palbociclib-treated NPC spheroids.
We have further assessed the cell cycle distribution of the NPC43 cells under different dosages of palbociclib from day 1 to day 5 (Fig. 1F). A sub-G1 peak indicating cellular apoptosis was only observed in NPC cells treated with high dose of palbociclib (20 µM). However, G1 arrest could be observed in NPC cells treated with lower doses of palbociclib. These results support that palbociclib has distinct effects on NPC cells at low and high doses. We have also performed RNA sequencing on the three NPC cell lines to identify the alterations of gene expression by NPC cell lines 24 hours post-palbociclib treatment. The differentially expressed genes were then subjected to KEGG database analysis (Fig. 1G). Gene expression in cell cycle progression was commonly downregulated in all three tested cell lines. The affected genes are listed in Table 1. The RNA expression pro ling again con rms that palbociclib causes cellcycle arrest pro le in NPC cells.

Oral administration of palbociclib suppressed growth of multiple NPC xenograft models in vivo
For comprehensive evaluation of the e cacy of palbociclib in NPC preclinical models, six xenografts representing early and advanced stages of NPC were included in this study. Xeno32, Xeno76, and C666-1 were established from the primary NPC, while Xeno23 and NPC43 were from recurrent NPC and C17 was originated from a metastatic NPC. The expression of p16, RB and cyclin D1 were examined in these xenografts in NOD/SCID mice by immunohistochemistry ( Supplementary Fig. S3). Expression of p16 was absent in all of these NPC xenografts. In contrast, RB and cyclin D1 were readily detected in all NPC xenografts. The palbociclib was prepared as suspension in deionized water. Mice bearing the NPC xenografts (Xeno32, Xeno76, C666-1, Xeno23, NPC43 and C17) were each fed daily with palbociclib (75 mg/kg/day) for 15, 23, 13, 54, 29 and 19 days respectively. The length of the treatment was dependent on the growth rate of NPC xenografts. The mice in the vehicle groups received an equivalent volume of deionized water on the treatment days. The tumor volume in each mouse was measured three times a week by a digital vernier caliper. Mice were euthanized at the end of the treatment period when the tumor size in the control group reached the diameter of around 1 cm. As shown in Fig. 2A -B, oral administration of palbociclib successfully inhibited the tumorigenic growth of all the NPC xenografts supporting the potential e cacy of palbociblib in treating NPC patients. We have also examined the expression of Ki-67, a commonly used cell proliferative marker in cancer, in NPC xenografts from control and treatment groups at the end of the study by immunocytochemistry (Fig. 2C-D). The quantitation process of the Ki-67 staining on sectioned tissues was illustrated in Supplementary Fig. S4. The percentages of Ki-67-expressing cells in the palbociclib treatment groups were all signi cantly lower compared to control groups in all the 6 NPC xenografts examined. The body weights of treated and control animals were measured throughout the treatment period, and no signi cance differences were observed suggesting that the doses of palbociclib used in the NPC-bearing mice had no major adverse effects on the general well-being of the animals (Fig. 2E).
The C666-1 cells used in this study were able to colonize in the lungs of the mice after tail vein injection. We have examined if palbociclib could suppress the lung metastasis of C666-1 cells in vivo. Five mice were included in each of the palbociclib treatment and control group. The duration of study lasted for 124 days. At the end of the study, lung tissues from the NOD/SCID mice injected with C666-1 cells were dissected, xed and processed for H&E staining to examine for the growth of C666-1 cells. The palbociclib was shown to effectively suppress the colonization of C666-1 cells in the lungs (Supplementary Fig. S5).
Combination treatment of palbociclib in NPC cells revealed antagonistic effects with cisplatin but synergistic effects with SAHA Drug resistance as well as side-effect due to high dose of cancer drug may worsen the treatment outcome in patients. Combination treatment with two or more anti-cancer drugs targeting different cellular pathways is a common strategy used to minimize treatment toxicity. We have examined the effects of combination treatment of NPC cells with palbociclib and other therapeutic agents including cisplatin or SAHA. Cisplatin is commonly used in clinical management of NPC in patients while SAHA has been evaluated in preclinical models of NPC (33). The action mechanisms of cisplatin and SAHA are distinct. The cisplatin is a DNA damaging agent while SAHA is known to be an effective histone deacetylation inhibitory agent. To evaluate if the combinational use of palbociclib with cisplatin or SAHA may have antagonistic, synergistic or additive effects, the NPC cells were treated in vitro either singly with palbociclib or in combination with cisplatin or SAHA at serially concentrations. In all the three tested NPC cell lines (NPC43, C17 and C666-1), the combined use of palbociclib plus cisplatin did not result in enhanced growth suppression of NPC cells, but instead revealed antagonistic effects comparing to treatment by cisplatin or palbociclib alone (Fig. 3A). This antagonistic interaction between palbociclib and cisplatin was not unexpected as the documented cytotoxic mechanism of cisplatin is on DNA damage which could only be elicited upon cell cycle entry. Palbociclib inhibits cell cycle entry which may protect the cells from the cytotoxic effect of cisplatin. In contrast, we observed that treatment of SAHA signi cantly enhanced the growth suppression effect of palbociclib on NPC cells (Fig. 3B). We have also employed the Chou-Talalay method to calculate the combination index (CI) of the drug combination which can indicate if the combination treatment of two reagents may have synergistic, additive or antagonistic effects (34,35). A value of CI less than one (CI < 1) implies synergistic effect. A CI value equal to one (CI = 1) implies additive effect while CI value more than 1 (CI > 1) implies antagonism (34,35). The CI at 2.5/2.5 µM (palbociclib/cisplatin) combination were 1.58, 2.54 and 1.041 respectively for NPC43, C17 and C666-1, indicating a signi cant antagonistic effect when palbociclib and cisplatin were used concomitantly to treat NPC cell lines (Fig. 3C). In contrast, the Chou-Talalay calculation supports a synergistic effect when palbociclib was used in combination with SAHA. The CI values at 1.0/0.1 µM (palbociclib/SAHA) combination for NPC43, C17 and C6661 were 0.36, 0.38 and 0.6 respectively supporting synergistic actions of combined treatment of palbociblib and SAHA (Fig. 3C).

Combined treatment of palbociclib and SAHA suppressed the growth of NPC xenografts in vivo
Based on the results of Chou-Talalay analysis (34,35), we carried further investigations to con rm the synergistic therapeutic effect of combination treatment of palbociclib and SAHA on tumorigenic growth of NPC grown in vivo using Xeno23, Xeno76 and C666-1. The NPC xenograft bearing mice were treated with palbociclib (75 mg/kg in alternate day) or/and SAHA (20 mg/kg in alternate day). The palbociclib dose was reduced from 75 mg/kg per day in the single-drug treatment to 75 mg/kg in alternate day in the combination treatment. A reduced dose of palbociclib was used to ensure, if any potential additive or synergistic effect of SAHA in such regimen could be observed. The results of the combination treatments are shown in Fig. 4A. The average tumor volume, growth curve, and the histological appearances of all NPC xenografts examined are shown. A signi cant inhibition of tumour growth was observed in all the combination treatment group (palbociclib + SAHA) compared to single treatment group (either palbociclib or SAHA). The decrease in tumour volume was particularly prominent in Xeno23 and C666-1. The expression of Ki-67 in NPC cells in the control and treatment groups was further examined for status of cell proliferation. The combination groups all revealed a signi cant decrease of Ki-67 expression when compared to their control groups in all the 3 NPC xenograft models (Fig. 4B).
We further conducted 18F-FDG micoPET scan for Xeno23 mice to examine the metabolic rate of the tumor after combination treatment. The metabolic rate of tumor was represented by SUVmean (standardized uptake value). The metabolic rates were 0.76 and 0.09 in the vehicle and combination treatment group (after normalized to liver basic metabolic rate) (Supplementary Fig. S6). The body weights of treated and control animals were measured throughout the treatment period, and no signi cance difference was observed (Fig. 4C).
Autophagy-associated cell death was involved in mediating the enhanced cytotoxicity of the combined treatment in NPC cells We then sought to explore some of the underlying mechanisms involved in the enhancement of cell death induced by the combination treatment of palbociclib and SAHA. Responses of NPC43 cells toward the treatment of palbociclib either alone or in combination with SAHA were examined ( Supplementary Fig. S7).
The expression levels of multiple markers in cell cycle, differentiation and apoptosis were analysed by Western blots. Treatment with palbociclib alone inhibited cell cycle progression as shown by the suppression of pRB-Ser780 and cyclin A expression in treated NPC43 cells. Treatment with SAHA alone at the dosages used was much less effective in suppressing RB phosphorylation and cyclin A expression indicating a different mode of action of SAHA from palbociclib. We did not observe signi cant impact of single or combination treatment of palbociclib and SAHA on involucrin (a squamous cell differentiation marker), cleaved-PARP and cleaved-caspase 3 (apoptosis markers) in NPC43 cells treated with SAHA alone or in combination with palbociclib.
To explore for potential biological pathways involved in the enhanced cell death of combination treatment of palbociclib and SAHA on NPC43 cells, we compared RNA-sequencing pro les of three NPC cell lines (C666-1, NPC43 and C17) to identify differential gene expression (DEG) after palbociclib + SAHA treatment. The Venn Diagram revealed 914 upregulated genes shared by three NPC cell lines (NPC43, C666-1 and C17 cells) treated by palbociblib + SAHA (Fig. 5A). KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analysis of these upregulated genes revealed the top 20 pathways based on the gene ratios related to each pathway. Interestingly, many of these upregulated genes are involved in autophagy pathways. The autophagy pathways were listed according to the CPDB (ConsensusPathDB) database (Fig. 5B). The autophagy-related genes involved in the enrich pathways of lysosome, macroautophagy,

Discussion
Over the past decade, targeted therapies are widely adopted and continuously investigated in various types of cancer. For many prevalent cancers such as gastric, breast, and liver cancers, there are plenty of FDA-approved targeted therapies based on the genomic properties of different tumors. Whereas in the treatment options of NPC, concurrent chemotherapy and adjuvant chemotherapy are still the main lines of treatments, especially for patients suffering from advance diseases (38). In the current phase III clinical trials of NPC, the majority of them are still assessing e cacy in treating NPC patients with combinational use of conventional cytotoxic drugs e.g. cisplatin, doxorubicin, 5-FU (39). The development of effective therapeutic agents against novel and selective molecular targets different from conventional cytotoxic drugs may provide enhanced therapeutic effectiveness to NPC treatment. At present, there are very few clinical trials involving the use of targeted drugs in NPC. Among them, cmrelizumab and tislelizumab, are used to target program cell death-1 (PD-1); anlotinib and nimotuzumab are used to target receptor tyrosine kinases and epidermal growth factor receptor (EGFR) respectively (40)(41)(42)(43). Their clinical effects are yet to be reported. mitophagy, autophagy are listed in Table 2. We further performed western blot analysis and revealed an increase of LC3-II (the phosphatidylethanolamine conjugated form of LC3)/LC3-I ratios which are indicative of an increase in autophagy ux in NPC cells after combination treatment by palbociclib and SAHA (Fig. 5C).
Elevated LC3-II/LC3-I ratios could be due to enhanced autophagosome formation or blockage of autophagic degradation. We then evaluated the effect of chloroquine (CQ), a speci c inhibitor to block autophagic ux by decreasing autophagosome-lysosome fusion, on the LC3-II/LCII-1 ratios and the cell viability. If the cell is undergoing a full autophagy process, CQ would induce an increased LC3-II accumulation in cells. If the autophagy process is blocked at the autophagic degradation step, CQ treatment would not further increase the expression levels of LC3-II (36). In Fig. 5D, the level of LC3-II in C666-1 cells in the combination treatment group was higher (increased by ~ 5.8 fold after normalized with GAPDH expression) compared to single treatment group by palbociclib (~ 2.9 folds) and SAHA (~ 2 folds) indicating increase of autophagy ux in cells treated by combination therapy. In the presence of autophagy ux inhibitor, CQ, the upregulation of LC3-II was further increased to 7.6 folds in C666-1 cells treated by combined palbociclib and SAHA. Importantly, when this autophagy ux was blocked by CQ, the viability of cells palbociclib in combination treatment cell increased signi cantly (from 31-50%) at 24 hours (Fig. 5E).The same trend was observed at 48 hours and 72 hours as well. The partial rescue of cell death after CQ addition to cells treated with palbociclib + SAHA suggests that autophagy was involved in the cell death of the combined treatment. We also examined the levels of LC3-II from protein extracted from C666-1 xenografts grown in mice after combination treatment with palbociclib and SAHA. A signi cant increase of LC3-II/LC3-I turnover ratio was also observed in the combination treatment group compared to the single drug treatment groups which agreed with the results of the Western blot analysis in C666-1 grown in vitro (Fig. 5F).

Palbociclib-resistant NPC cells were sensitive to cisplatin treatment
As it is common for cancer cells to develop mechanisms to resist the inhibitory effect of a particular drug after prolonged treatment, NPC cells may eventually gain resistance to palbociclib. To provide preclinical evidence in treating palbociclib-resistant NPC, we carried investigation to examine if palbociclib-resistant NPC cells retain their responsiveness to other chemotherapeutic agents such as cisplatin. We were able to establish palbociclib-resistant (PD-R) NPC cells lines from C666-1 and NPC43 by treating them with an increasing dose of palbociclib over a period of one and a half year. We then characterized the expressions of cell cycle related proteins in the parental and resistant NPC43 under the vehicle and palbociclib treatments (Fig. 6A). RB and pRB-Ser780 were decreased in NPC43 PD_R cells. Notably, the basal expression of cyclin A was still maintained in resistant cells but could not be downregulated by 5 µM of palbociclib, which is the concentration that can effectively inhibit the cyclin A expression in the parental NPC43 cells. The expressions of cyclin E1, cyclin D2, and cyclin D3 were higher in the resistant NPC43 cells. All these results indicate the resistant cells may have acquired alternate pathways, which are independent of CDK4/6/cyclin D1/RB pathway, to maintain cell proliferation under the presence of palbociclib. Moreover, the resistant lines also had a downregulated E-cadherin expression, but an upregulated N-cadherin expression, suggesting the resistant line may be more prone to metastasis. We also examined the expression of the cancer stemness related genes. MMP2, MMP9, Nanog, and SOX2 mRNA were identi ed to be elevated through the qPCR test (Fig. 6B). These results suggested that palbociclib-resistant NPC43 cells may acquire an increase in cancer stemness.
We have conducted colony formation to verify the resistance of palbociclib in NPC43 (Fig. 6C). The parental NPC43 could no longer sustain the formation of colonies under the treatment of 0.8 µM palbociclib, while the NPC43 PD_R cells could maintain colony growth under the treatment of 8 µM palbociclib.
We then examined the response of the two PD-resistant NPC cell lines to the treatment of cisplatin (Fig. 6E). In NPC43, the IC50 day5 of cisplatin in the palbociclib-resistant cells (1.66 µM) was slightly lower than that of the parental cell (2.32 µM), indicating the resistant line was even more sensitive to cisplatin than the parental line. Besides, the IC50 day5 of cisplatin treatment of C666-1 parental and C666-1 PD_R cells was similar. This suggests that cisplatin can potentially be used in treating palbociclib-resistant NPC patients.

Cisplatin-resistant NPC cells were sensitive to palbociclib treatment
Since platinum-based therapies have been being utilized in rst-line treatment of primary NPC, salvage treatment in recurrence as well as palliative treatment of metastasis (2,3,37), we were interested to assess if cisplatin-resistance NPC cells were responsive to palbociclib treatment. We have obtained two cisplatin-resistant sublines of NPC43 which showed a 6-fold increase of IC50 compared to the parental NPC43 (IC50 day2 increase from 6.8 uM to 48.58 uM and 51.31 uM for two sublines) (Fig. 6F). We then determined the e cacy of palbociclib in treating these cisplatin-resistant sublines. The IC50 day3 of palbociclib for NPC43 parental and cisplatin-resistant sublines were 28.61 µM versus 19.74 µM and 16.83 µM (Fig. 6G). The slight decrease of IC50 to palbociclib indicated the cisplatin-resistant NPC cells had similar sensitivity to palbociclib as their parental cells.
In this study, we aim to provide preclinical evidence to develop the incorporation of palbociclib as another targeted drug for NPC and assess the compatibility of palbociclib with other chemotherapies in NPC treatment. Recent genomic pro ling of NPC disclosed that druggable targets are uncommon in NPC (10,11,13,14). Nevertheless, a dysregulated p16-CDK4/6-cyclinD1 signaling has long been identi ed as a common event in the pathogenesis of NPC involved in promoting uncontrolled growth and metastasis of NPC.
We have used a large panel of NPC preclinical models including our newly established and well-characterized xenografts and cell lines derived from primary, recurrent and metastatic NPC (44) to evaluate the tumor-suppressive effect of palbociclib. We con rmed that functional RB and inactivated p16 are common features in our tested NPC models as described in our published study (8). The sensitivity of NPC to palbociclib was con rmed in vitro and in vivo in NPC cell lines and xenografts by arresting the NPC cells in G1 phase of cell cycle. All NPC xenografts examined were sensitive to a daily dose of 75 mg/kg/day, which is comparatively lower to the doses used in many reported in vivo studies in other cancer types (150 mg/kg/day to 200 mg/kg/day) (45)(46)(47). The responses of different NPC xenografts to palbociclib were variable ( Supplementary Fig. S8), which may be related to the varied basal expression of some of the sensitive markers to palbociclib or other intrinsic properties of various NPC cell lines and NPC xenografts. Among the NPC xenografts examined, the Xeno23, Xeno76 and Xeno32 had the best response, as the growth of tumors was arrested during the entire treatment periods (Fig. 2B). NPC xenografts from established NPC cell lines (NPC43, C666-1, and C17) had a slight increase in tumor size during the treatment periods but the tumour sizes were still signi cantly smaller in treatment groups compared to the vehicle control groups. As an FDA-approved drug for clinical use, the side effect of palbociclib should not be a signi cant issue. In this preclinical trial, we did not observe signi cant difference in body weights between mice in the treatment and control groups. Interestingly, we also observed that palbociclib could inhibit the lung colonization of NPC cells in vivo after tail-vein injection suggesting its potential inhibitory effect in NPC metastasis ( Supplementary Fig. S5). In breast cancer, palbociclib was shown to inhibit metastasis in animal models through inhibition of the c-Jun/COX-2 signaling pathway (48). The detail molecular mechanism of how palbociclib suppresses metastasis of NPC warrants more investigation.
The potent suppressive effect of palbociclib on growth of NPC xenografts derived from NPC patients with primary, recurrent and metastatic cancer supports its application in the clinical trials of NPC treatment. A previous preclinical study using another CDK4/6 inhibitor, ribociclib, which interferes a different site in the ATP-binding pocket of CDK4/6 as that of palbociclib, also demonstrated the inhibition of CDK4/6 could inhibit the growth of NPC cells (49). An integrated genomic and transcriptomic study of 5 patient-derived xenografts also discovered the copy number of CCND1 and CDKN2A are the potential drug target of palbociclib to suppress tumor growth (50). There was also a case report demonstrating the clinical bene ts from palbociclib in a patient with previously treated metastatic NPC with CDK4 ampli cation (51). All these observations indicate that targeting the dependency of NPC cells in this cell cycle signaling pathway is of high therapeutic value to NPC patients.
In the second part of this study, we took a forward step in exploring the e cacy of combination treatments with palbociclib. A combined regimen of chemodrugs in cancer treatment is common to prevent the development of drug resistance, reduce drug dosage and minimize side effects of the drug. Since platinum-based chemotherapy (e.g. cisplatin) is the most commonly used treatment of NPC, we rst assessed the combined effect of cisplatin and palbociclib in affecting the viability of NPC cells in vitro. Previous studies have reported an antagonistic effect of CDK4/6 inhibitor against other chemodrugs (52,53). Robust antagonism to CDK4/6 inhibitor was observed with combination use of speci c chemotherapeutic drugs, such as taxane, PLK1 inhibitors, gemcitabine, and other drugs, the actions of which are dependent on ongoing cell cycle progression. In this study, palbociclib also protected the cells from the cisplatin-induced cytotoxic effect in NPC cells (Fig. 3A). This may be explained by the action of palbociclib to induce cell cycle arrest in the G1 phase. The cytotoxic action of cisplatin is to induce DNA damage by crosslinking purine bases on DNA, interfering DNA damage repair, and triggering apoptosis and cell death (54). These cytotoxic actions of cisplatin take place in the S phase and may be abrogated by the cell cycle inhibitory property of palbociclib.
On the other hand, we observed a synergistic effect of palbociclib with SAHA, which is an FDA-approved drug for cutaneous T cell lymphoma (55). SAHA is a broad-spectrum HDACi which disables the HDAC in removing the acetyl groups from the protein histone and thus disrupting the normal regulation on gene expression (56,57). SAHA can cause growth arrest and death of a broad variety of transformed cells both in vitro and in vivo at concentrations that have little or no toxic effects on normal cells (23,58). The anticancer effect of SAHA is believed to be related to dysregulating gene expression involved in cell proliferation and cell death pathways (23). The exact mechanisms of anticancer property of SAHA remains to be further elucidated.
We con rmed the synergistic effects of palbociclib and SAHA in three authenticated EBV-infected NPC cell lines (Fig. 3B). Moreover, enhanced tumor inhibition effect by the palbociclib + SAHA combination treatment was also observed in Xeno76, Xeno23, and C666-1 in mice models (Fig. 4A). This is the rst report demonstrating a concurrent use of SAHA in enhancing the therapeutic e cacy of palbociclib in NPC. We sought to understand the mechanisms behind the enhanced cell death in this combination treatment. In a previous study, SAHA has been reported to induce apoptosis and suppress tumor growth in NPC through the activation of the lytic cycle of EBV (33,59). By doing RNAscope of lytic gene, the lytic reactivation of EBV could be revealed at single-cell level ( Supplementary Fig. S9). However, only less than 1% of cells in the tumors of mice were shown to express EBV lytic gene (BZLF1), suggesting the lytic reactivation was not the major cause for the enhanced cell death in the combined treatment. Upregulation of apoptosis and differentiation markers was also not detected in the cells under combined treatment ( Supplementary Fig. S7), suggesting the enhanced cell death was not due to promoted apoptosis or differentiation. Interestingly, through RNA sequencing and western blot analyses, we found autophagy-related pathways were commonly induced in the combination treatment of 3 NPC cell lines (Fig. 5). There have been evidences showing that autophagy can participate in a caspase-independent form of programmed cell death induced by anticancer drugs (60). This type of cell death is due to the accumulation of presumably toxic autophagic cargo in cells which are defective in the lysosomal capacity to degrade this material. Autophagy has been reported to play contradictory roles in tumor initiation and progression. Both repression and stimulation of autophagy have been indicated as therapeutic approaches depending on the cell context of the tumors. Therapeutic induction of autophagy-associated cell death can be accomplished by modulation of regulators of autophagy. Combining autophagy associated mTOR inhibition with radiation was reported to enhance therapeutic effects in cancer cells and xenografted tumors (61). In our study, palbociclib could elevate the expression of LC3-II, and the combined use of palbociclib + SAHA could further augment the LC3-II elevation (Fig. 5). CQ is an autophagy inhibitor that can alkalize lysosome lumen and block autolysosome degradation. We found that CQ treatment further enhanced the increased expression level of LC3-II in both palbociclib and palbociblib + SAHA treatment groups, indicating that palbociclib alone or in combination with SAHA could upregulate autophagy ux. Furthermore, we found that the tumor cell death induced by palbociclib or combination treatment could be reversed by CQ in C666-1. In the in vivo study, the combined treatment also induced a higher LC3-II/LC3-I ratio in the mice tumors (Fig. 5F). All these suggest that autophagy is associated with the cell death induced in cells treated with palbociclib alone or in combination with SAHA. A previous study showed that palbociclib induced autophagy in hepatocellular carcinoma cells in a CDK4/6 independent manner, but through a mechanism involving 5′ AMP-activated protein kinase (AMPK) activation and protein phosphatase 5 (PP5) inhibition (62). Some earlier studies reported HDAC inhibitors could induce caspase-independent autophagic cell death in HeLa and chondrosarcoma (63,64). In our NPC cell systems, SAHA seemed to potentiate the autophagic induction effect of palbociclib. This may lead to massive degradation of essential cellular structures and thereby causing autophagy-associated cell death. Investigations will be required to further elucidate the role of autophagy in mediating the enhanced cytotoxic effects of palbociclib used either alone or in combination with SAHA especially in the cell context of NPC.
Our study has demonstrated the potent effect of palbociclib in suppressing the tumor growth of primary, recurrent and metastatic cancer cells. First-in-human clinical trial of palbociclib in NPC patients will be anticipated. In 2009, 16 NPC patients with treatment-naïve WHO III histological grade, stage II (n = 2), III (n = 6), and IVB (n = 8) disease was treated with non-selective CDK inhibitor, seliciclib (50). The later development of speci c inhibitor to CDK4/6 including palbociclib displaced the use of seliciclib in the clinical trials probably due to less side effects. Considering the use of palbociclib may eventually lead to emergence of drug resistance (65,66), we aim to examine if the palbociclib-resistance NPC cells would still be vulnerable to cisplatin treatment. We rst established C666-1 and NPC43 sublines which are palbociclib resistant (C666-1 PD_R and NPC43 PD_R). By examining the protein expression involved in regulation of cell cycle progression, acquisition of new cell cycle pathway independent of phosphorylation of the RB protein was observed (Fig. 6A). With the loss of total RB and pRB-Ser780, the cyclin A continued to be expressed at high level which indicates the proliferation ability of palbociclib-resistant NPC43 cells under palbociclib treatment. Overexpression of cyclin E1 was also observed in palbociclib-resistant NPC43 cells. Activation of E2F through the cyclin E-CDK2 axis is believed to be involved to overcome the inhibition of CDK4/6 for cell cycle progression from G1 to S phase (66). NPC cells may employ similar mechanism to develop resistance to palbociclib. Importantly, cisplatin is still effective to these palbociclib-resistant cell lines (Fig. 6E), suggesting that cisplatin can still be used to treat NPC patients who have developed resistance to palbociclib.
In spite of the promising prospect of utilizing palbociclib as a therapeutic option of NPC treatment, irradiation and concurrent cisplatin treatment will remain as the rst-line treatment of most primary cases until the bene ts of palbociclib could be demonstrated with clinical trials enrolled with recurrent or metastatic NPC patients. Moreover, due to the lack of approved targeted therapy to NPC, cisplatin is used in chemo-regimens of both salvage treatment of recurrent tumors and even palliative treatment of metastases (3). Along this, we sought to verify if palbociclib could be used to treat cell lines resistant to cisplatin. Palbociclib was found to be effective to suppress the growth of cisplatin-resistant NPC43 cells (Fig. 6G). The cisplatin and palbociclib could potentially be used alternatively to patients who have developed drug resistance to the other drug.

Conclusion
In conclusion, palbociclib effectively induced cell cycle arrest and suppressed the growth of NPC in models derived from primary, recurrent and metastatic NPC. The combined use of SAHA further potentiated the inhibitory effect of palbociclib. Concurrent use of cisplatin and palbociclib is not advised. NPC cells which developed tolerance of palbociclib remain sensitive to cisplatin, and vice versa. Together, the work herein provides relevant information for planning clinical application of palbociclib-involved regimens in NPC treatment.      p<0.0001 ****, p<0.001***, p<0.005**, p<0.01*.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementaryFigures.pdf