The PI3K/AKT/mTOR signaling pathway is aberrantly activated in primary central nervous system lymphoma and correlated with a poor prognosis

Tumor samples were acquired by surgical resection or stereotactic biopsy of PCNSL in 43 immunocompetent patients who were newly diagnosed and treated at Beijing Tiantan Hospital, Capital Medical University from December 2015 to February 2020. All 43 patient specimens were fixed with formalin. Since most tumor specimens were obtained by stereotactic biopsy, only 7 patients were able to have excess tissue specimens stored at − 80 °C. All the patients met the following criteria: 1) age over 18; 2) no glucocorticoid treatment before the tumor tissues were obtained; 3) at least 3 cycles of high-dose methotrexate-based induction chemotherapy after diagnosis; 4) no acquired and congenital immunosuppression situation. The patients were followed up until December 2020. Clinical characteristics, including age, sex, Eastern Cooperative Oncology Group (ECOG) score, lactate dehydrogenase (LDH) at initial diagnosis, number of lesions, deep brain involvement, pathological subtypes, recurrence, and survival outcome, were collected for all 43 patients. The progression-free survival (PFS) and overall survival (OS) of the patients were assessed. In addition, 6 samples of reactive hyperplastic lymph nodes were collected randomly as the control group.

The tissue samples were fixed with formalin and embedded in paraffin to make blocks. These paraffin-embedded samples were cut into 4-μm sections. After dewaxing with xylene and rehydration with graded ethanol, the sections were heated for antigen retrieval in citrate buffer (10 mM, pH 6.0) in a microwave. Then, 0.3% hydrogen peroxide was used to inactivate endogenous peroxidases, and 10% goat serum was used to block the nonspecific binding sites. After washing with PBS, sections were incubated overnight at 4 °C with the following primary antibodies: p-AKT^Ser473 (1:100 dilution, Cell Signaling Technology (CST), Beverly, MA, USA), p-mTOR^Ser2448 (1:50 dilution, Abcam, Cambridge, UK), p-S6^Ser240/244 (1:2000 dilution, CST) and p-4E-BP1^Thr37/46 (1:1600 dilution, CST). The next day, these sections were incubated with the working solution of HRP-linked secondary antibody at room temperature for 1 h and then developed with 3,3′-diaminobenzidine (DAB) solution. Finally, the sections were counterstained with hematoxylin. The slides were digitally scanned using Vectra Polaris slide scanner (PerkinElmer) at 40x magnification. The whole slide images were processed using Phenochart software. Two independent pathologists, who were blind to the clinical data, evaluated the staining results from five randomly selected high-power fields.

Positive staining was assessed according to the intensity and the range of positive staining. The intensity was scored from 0 to 2 (0 for absent staining, 1 for weak staining, and 2 for moderate or strong staining). The range was indicated by the percentage of positive staining, scored from 1 to 3 (1 for < 10%, 2 for 10–50%, 3 for > 50%) [30]. Generally, protein expression was considered positive if the product of the intensity and range scores was greater than 3.

Western blotting was performed on frozen tissue samples preserved at − 80 °C. We used RIPA lysis buffer, proteinase inhibitors and phosphatase inhibitors as the protein extraction reagents. A bicinchoninic acid (BCA) protein analysis kit was employed to examine the protein concentration of each sample. The proteins of the samples were loaded onto sodium dodecyl sulfate (SDS) polyacrylamide gels and transferred to polyvinylidene fluoride (PVDF) membranes after electrophoresis. In general, the membranes were subsequently blocked by using 5% skimmed milk and incubated for 1 h at room temperature under agitation. For phosphorylated protein detection, we used 5% bovine serum albumin (BSA) for blocking. Then, the membranes were incubated overnight at 4 °C with the following primary antibodies: pan-AKT (CST), p-AKT^Ser473 (CST), m-TOR (CST), p-mTOR^Ser2448 (Abcam), S6 (CST), p-S6^Ser240/244 (CST), 4E-BP1 (CST) and p-4E-BP1^Thr37/46 (CST) at a dilution of 1:1000. β-Actin was used as an internal reference. After that, the membrane was incubated with HRP-linked secondary antibodies at room temperature for 1 h. Finally, the membrane was exposed to film using GBox-Chemi XX9 (Syngene) after being incubated with enhanced chemiluminescence substrate. We used GeneTools software to evaluate the grayscale intensity and perform quantitative analysis of the extracted protein.

Total RNA was isolated using TRIzol and converted to cDNA using the Transcriptor First Strand cDNA Synthesis Kit (Roche) according to the manufacturer’s protocol. The DNA primers for the AKT1, MTOR, RPS6 and EIF4EBP1 genes were designed with Primer-BLAST software, and the sequences of each primer are listed in Supplementary Table S1. We chose PowerUp SYBR Green Master Mix (Applied Biosystems) as the reagent. qPCR assays were run on 7500 Fast Real-Time PCR System (Applied Biosystem). The relative expression levels of each target gene were determined by the 2^-△△CT method.

In this study, we conducted FISH to detect the deletion of the PTEN gene in PCNSL samples. Formalin-fixed, paraffin-embedded (FFPE) tissue blocks were sectioned at 4-μm thickness, deparaffinized, and rehydrated in xylene and graded ethanol. Subsequently, the slides were placed in the pretreatment reagent (Abbott Vysis) for tissue digestion for approximately 45 min. The digestion time was adjusted appropriately according to the degree of tissue digestion. The slides were then mixed with PTEN (10q23) probe (Abbott Vysis) according to the instructions of the manufacturer, placed into a hybridization system (Abbott Molecular Thermobrite Stat Spin), denatured for 20 min at 75 °C and incubated overnight at 37 °C. After washing the slides with saline-sodium citrate (SSC) buffer and IGEPAL (Sigma) to remove nonspecific background signals, the slides were counterstained with diamidino-2-phenylindole (DAPI) for fluorescence microscopy observation. The slides were scanned using LSM-880 laser-scanning confocal microscope (Zeiss). The images were processed with ZEN software. Five randomly selected high-magnification fields were used to analyze at least 100 interphase nuclei for each PCNSL sample. The PTEN gene was considered deleted in the sample if the ratio of the gene probe to the centromere probe was less than 0.8 [31].

Statistical analysis was performed with SPSS 26.0 software. Two-tailed t-tests were used for parameter analysis. The frequency analysis of the nominal data was performed by Pearson’s χ² test or two-tailed Fisher’s exact test. The Pearson correlation coefficient or Spearman rank correlation coefficient was used for correlation analysis. Kaplan-Meier curves were used for survival analysis, and the log-rank test was used to determine the significance for survival comparisons. The prognostic factors were analyzed by a multivariate Cox proportional hazards regression model. Differences with P < 0.05 were considered statistically significant.

Immunohistochemical staining was performed on 43 PCNSL tissue specimens and 6 reactive hyperplastic lymph node specimens (the control group). The staining results showed positive staining in brown or brownish-yellow colors. For the p-AKT and p-mTOR proteins, positive staining was mainly located in the cytoplasm. The p-S6 protein was mainly expressed around the nucleus. The p-4E-BP1 protein was mainly expressed in both the nucleus and cytoplasm. The IHC results of the above proteins are shown in Fig. 1.

For p-mTOR, the positive expression rate was 72.1% (31/43) in the PCNSL tissue samples versus 16.7% (1/6) in the control samples, with a P value of 0.027. The only lymph node sample with p-mTOR positivity was the same sample that expressed p-AKT, as mentioned above.

In addition to IHC, we used Western blotting to further confirm the expression of the abovementioned phosphorylated proteins in PCNSL. The results are shown in Fig. 2. The grayscale values of each sample band after internal reference correction were measured and calculated. The differences in the expression levels of each targeted protein between the PCNSL group (n = 7) and the control group (n = 6) of samples were analyzed by t-tests. The results indicated that the expression levels of p-mTOR, p-S6 and p-4E-BP1 in the PCNSL samples were significantly higher than those in the reactive hyperplastic lymph node samples, and the differences were statistically significant (Table 1). The results were roughly consistent with the results of IHC.

In conclusion, the IHC and Western blotting results demonstrated that p-AKT, p-mTOR, p-S6 and p-4E-BP1 were overexpressed in the PCNSL group compared to the reactive hyperplastic lymph node group. Since these proteins are all key molecules in the PI3K/AKT/mTOR signaling pathway, we speculated that this pathway is abnormally activated in PCNSL.

The gray values of total protein and phosphorylated protein of mTOR, AKT, S6 and 4E-BP1 in 7 PCNSL samples were measured simultaneously by Western blotting. The ratio of the grayscale value of the phosphorylated protein to that of the total protein was calculated to evaluate the phosphorylation level of each protein. Pearson correlation coefficients were used to analyze the relationship of the phosphorylation level of mTOR with that of AKT, S6, and 4E-BP1. The results showed that the correlation coefficient between the phosphorylation level of mTOR and S6 was 0.871 with a P value of 0.011, which was statistically significant. However, there was no significant correlation between mTOR phosphorylation level and the other two proteins (P > 0.05).

We selected the following related genes, AKT1, MTOR, RPS6 and EIF4EBP1, and detected their mRNA expression in 5 samples of PCNSL by real-time qPCR. Three samples of reactive hyperplastic lymph nodes were used as the control group. The relative mRNA expression levels of these target genes are shown in Table 2. The results showed that the relative mRNA expression level of MTOR in PCNSL samples was 6.363 ± 2.864, which was significantly higher than the 1.089 ± 0.503 value in the control group; the P value was 0.013 (Fig. 3).

The status of PTEN loss was detected by FISH in 43 FFPE PCNSL samples. Since the staining results of 6 samples were unsatisfactory, a total of 37 samples were available for analysis. The rate of PTEN loss was 18.9% (7/37) (Fig. 4). According to the results of PTEN detection, 37 samples of PCNSLs were divided into two groups: the PTEN loss group and the normal group. The positive rates of p-AKT, p-mTOR, p-S6 and p-4E-BP1 expression from the immunohistochemical analysis in the two groups are shown in Table 3. The positive rate of p-AKT in the PTEN loss group was 100% (7/7), while in the normal group, the positive rate was only 56.7% (17/30). The difference was statistically significant (P = 0.038). However, there was no significant difference in the positive rates of other target proteins between the two groups. We used the Spearman rank correlation coefficient to analyze the relationship between PTEN loss and p-AKT expression. The results showed that the correlation coefficient was 0.356 and that the P value was 0.031, suggesting that PTEN loss was correlated with p-AKT expression.

We collected the clinical characteristics of 43 patients with PCNSL. The median age was 59 years old (range from 16 to 78), and the proportions of males and females were similar (53.5% vs. 46.5%). All the cases were pathologically confirmed as DLBCL, and approximately 83.7% of them were of the nongerminal center B-cell like (non-GCB) subtype. The ECOG score was ≥2 in the vast majority (93.0%) of patients. However, elevated LDH was found in only 7 patients at initial diagnosis. A total of 62.8% of the patients had single lesions, and 74.4% had deep brain involvement, which referred to lesions involving the cerebellum, brainstem, corpus callosum or basal ganglia. After at least 3 cycles of high-dose methotrexate-based induction chemotherapy, the objective response rate (ORR) was 79.1%, and the complete remission (CR) rate was 37.2%. By the last follow-up (median time 18 months), approximately half (48.8%) of the patients had relapsed, and 6 patients had died.

According to the above clinical characteristics, we divided the PCNSL patients into different groups. The protein expression levels of p-AKT, p-mTOR, p-S6, p-4E-BP1 and the loss of PTEN in each group are shown in Supplementary Table S2. Since FISH results were only available for 37 patients, the analysis of PTEN loss was performed using clinical data from these 37 patients. The results indicated that the rate of relapse in the p-mTOR-positive group was 64.5% (20/31), while that in the p-mTOR-negative group was only 8.3% (1/12). Similarly, the rate of relapse in the p-S6-positive group was significantly higher than that in the p-S6-negative group (58.8% vs 11.1%). The P values were 0.001 and 0.021, respectively. We also found that the rate of PTEN loss was 66.7% (4/6) in patients with the GCB subtype and only 9.7% (3/31) in patients with the non-GCB subtype, and the difference was significant (P = 0.007). There was no significant correlation between the other targets and the clinical characteristics.

We evaluated the prognosis of PCNSL in terms of OS and PFS. By the end of follow-up, only 6 (14.0%) of the 43 patients had died, so the median OS was not reached. The median PFS calculated by the Kaplan-Meier method was 23 months (95% CI 13.9–32.1 months). The estimated 2-year PFS rate was 40.2%, and the 3-year OS rate was 76.8%. The Kaplan-Meier survival curves for OS and PFS were drawn according to the protein expression levels of p-AKT, p-mTOR, p-S6, and p-4E-BP1 and the loss of PTEN, which are shown in Fig. 5. The PFS of the patients with positive p-mTOR or p-S6 expression was significantly shorter than that of the patients with negative expression of the respective proteins, and the estimated 2-year PFS rates were 23.3% vs 97.1 and 26.1% vs 88.9%, respectively, with P values of 0.002 and 0.009. (Fig. 5C, Fig. 5E) In addition, Kaplan-Meier analysis also indicated that the OS of the PTEN loss group was lower than that of the PTEN normal group, with a P value of 0.072, close to the statistical significance threshold.

Cox multivariate regression analysis was used to explore the independent risk factors affecting OS and PFS in patients with PCNSL. The including risk factors were as follows: age, sex, ECOG score, LDH level, number of lesions, whether deep brain involvement, pathological subtypes, relapse situation, the protein expression of p-AKT, p-mTOR, p-S6, and p-4E-BP1 and the loss of PTEN. The results showed that p-mTOR expression was an independent prognostic factor for a shorter PFS with a hazard ratio (HR) of 7.849 (95% CI: 1.037–59.408) and a P value of 0.046.