Serum ferritin level is an effective prognostic factor for lung cancer immunotherapy Authors: Fei Wang, Guodong Deng, Ning Liang, Pingping Hu, Kuo Liu, Tong Liu, Yang Li, Meng Yuan, Li Liu, Jian Xie, Lili Qiao, Fengjun Liu, Jiandong Zhang Published: 2023 ABSTRACT ABSTRACTImmunotherapy of lung cancer has achieved promising clinical results. However, it is urgent to develop predictive biomarkers for effective immunotherapy. While ferroptosis plays a critical role in immunotherapy efficacy, ferritin is an important regulatory factor. We, therefore, hypothesize that basal serum ferritin levels before immunotherapy and their corresponding changes during immunotherapy can be useful predictors of immunotherapy response in patients with lung cancer. We measured serum ferritin levels in 107 patients with lung cancer before and during immune checkpoint blockade treatments and studied the correlation between ferritin levels, response rate, and survival. Moreover, the correlation between basal ferritin and PD-L1 expression, tumor stages and pathological types was also analyzed. Patients with lower basal serum ferritin levels before immunotherapy had longer progression-free survival (PFS) (median 7 vs 4 months, P = .023) and higher disease control rate (DCR) (X2 = 4.837, P = .028), those with downregulated serum ferritin levels during immunotherapy correlated with longer PFS (median 9.5 vs 4 months, P < .001) and higher DCR (X2 = 6.475, P = .011). However, the “integrated factor”, which was calculated as the combination of lower basal serum ferritin levels before immunotherapy and downregulated serum ferritin levels during immunotherapy, correlated with prolonged PFS (P < .001). Multivariate analyses revealed that the basal serum ferritin levels before immunotherapy and the corresponding changes during immunotherapy were both strong independent prognostic factors (hazard ratio (HR) = 1.60, P = .041; HR = 2.65, P = .001). These findings suggest that serum ferritin levels can be used as a prognostic biomarker for lung cancer in predicting immunotherapy efficacy. Introduction According to statistics, lung cancer is the most common and lethal type of malignancy.1,2 Every year, over two million people globally are diagnosed with lung cancer and 1.5 million deaths are attributable to lung cancer.3,4 With numerous advances made in the past decade, immunotherapy is now regarded as a revolution in the history of tumor therapy, following traditional therapy. Immunotherapy of lung cancer has achieved promising results in clinical trials.5 For instance, in a randomized phase III study comparing nivolumab and docetaxel in advanced squamous non-small cell lung cancer (NSCLC) involving 272 patients, the median overall survival was 9.2 months with nivolumab versus 6.0 months with docetaxel, while the median progression-free survival (PFS) was 3.5 months with nivolumab versus 2.8 months with docetaxel (hazard ratio (HR) for death or disease progression, 0.62; 95% confidence interval (CI), 0.47–0.81; P < .001).6 Therefore, for patients without targeted genetic alterations or contraindications to programmed cell death protein 1/ligand 1 (PD-1/PD-L1) inhibitors, a combination therapy involving two or more immunotherapies has become the first-line standard of care for advanced squamous and non-squamous lung cancers.7 Certainly, biomarkers are required to guide treatment in immunotherapy. Currently, the biomarkers recommended in clinical practice guidelines to predict the efficacy of immunotherapy drugs include PD-L1, microsatellite instability (MSI), and tumor mutation burden (TMB). PD-L1 is the most commonly used biomarker in lung cancer. However, many patients with lung cancer having high PD-L1 expression do not respond to immunotherapy or see no long-term clinical benefits,8 and the cumulative side effects may be fatal. We, therefore, need better biomarkers for predicting the efficacy of immunotherapy in lung cancer. Unlike other forms of cell death, ferroptosis manifests as lipid peroxidation. Recent studies have found that ferroptosis is one of the most important targets of immunotherapy. According to a study, interferon-gamma (IFNγ) released by CD8+ T cells was found to downregulate the expression of SLC3A2 and SLC7A11, two subunits of the glutamate-cystine reversed transport system xc-, impairing cystine uptake by tumor cells, thus promoting lipid peroxidation and ferroptosis.9 Researchers have found that macrophages can clear ferroptotic cells, which proves that the immune system can clear tumor cells via the ferroptosis pathway.10 Furthermore, ferroptosis was found to emit chemotactic signaling and promote immune cell activation, suggesting that it is a more immunogenic type of cell death unlike apoptosis, and has the potential for effective combination immunotherapy.11,12 Therefore, ferroptosis plays a critical role in cancer immunotherapy, and the associated factors might be useful in predicting the efficacy of treatment in immunotherapy. Ferritin, a 24-subunit protein, has a large icosahedral cage-like structure with a molecular weight of ~ 450 kDa and inner and outer diameters of 8 nm and 12 nm, respectively.13,14 It is made up of the H- and L-subunits.15 Ferritin is a key regulator of the intracellular labile iron pool (LIP), which is required for ferroptosis. Studies have shown that reduced ferritin favors ferroptosis. The increase in LIP during ferritin catabolism promotes the accumulation of reactive oxygen species (ROS) and leads to ferroptosis.16 Scientists have found that the knockdown of ferritin heavy chain may increase free iron, making HA22T/VGH (human-derived hepatoma cell line) more sensitive to RSL3 ((1S,3 R)-RSL3, ferroptosis inducer) and iron-induced ferroptosis, and inducing Fenton chemically-associated ferroptosis.17,18 Some antitumor drugs that regulate REDOX levels can induce ferroptosis in tumor cells by regulating the iron pool. Dihydroartemisinin, for example, promotes ferritin degradation, increases the iron pool, accelerates ROS accumulation, and induces ferroptosis in cancer cells by regulating the AMPK/mTOR/p70S6k signaling pathway.19,20 In addition, inhibition of nuclear receptor coactivator 4 can inhibit ferritin degradation and ferroptosis, while its overexpression has the opposite effect.16 These findings indicate that lowering ferritin levels can promote ferroptosis. Because ferroptosis is an important component of the immunotherapy pathway, we hypothesize that ferritin might be an effective predictor for selecting patients for immunotherapy as a PD-L1 does. Patients with low ferritin levels should have higher free iron levels and respond better to immunotherapy. In this study, we measured the serum ferritin levels in patients with lung cancer before and during immune checkpoint blockade (ICB) treatments to see if it can effectively predict the efficacy of tumor immunotherapy. Furthermore, the study also looked at the relationship between serum ferritin level and PD-L1 expression, tumor stages and pathological types. Results Patients A total of 107 patients (83 men and 24 women) were enrolled in this study, of which 47 patients were ≥65 years old and 60 patients were <65 years old, and the median age was 64 (range: 39–79). The number of patients with small cell lung cancer (SCLC), lung squamous cell carcinoma (LUSC), and non-small non-squamous cell carcinoma (NS-NSCLC) were 28, 20, and 59, respectively. There were 28 patients with stage I-III disease and 79 patients with stage IV disease. Eighty-two patients accepted immunotherapy combined with chemotherapy, 30 patients accepted immunotherapy combined with targeted therapy, and 20 patients accepted immunotherapy combined with radiotherapy. Thirty-one patients agreed to surgery, while the remaining 76 did not. While 36 patients experienced immunotherapy-related side effects, 71 patients did not. Table 1 shows the baseline patient characteristics. Table 2 summarizes the immunodrugs used in this study.Table 1.Demographic and clinicopathological characteristics of patients.Patient CharacteristicsNo. (%) (N = 107)Sex Male83 (77.57)Female24 (22.43)Age(years) <6560 (56.07)≥6547 (43.93)Mean Age64 years oldStage I-III28 (26.17)IV79 (73.83)Histopathology SCLC28 (26.17)LUSC20 (18.69)NS-NSCLC59 (55.14)Treatment Combined with chemotherapy82 (76.64)Combined with targeted therapy30 (28.03)Combined with radiotherapy20 (18.69)Surgery Yes31 (28.97)No76 (71.03)Side effects of immunotherapy Yes36 (33.64)No71 (66.36)Table 2.The number and proportion of different immunodrugs.ImmunodrugsNo. (%) (N = 107)Tislelizumab20 (18.69)Pembrolizumab22 (20.56)Sintilimab20 (18.69)Camrelizumab22 (20.56)Toripalimab5 (4.67)Durvalumab14 (13.08)Atezolizumab1 (0.93)Nivolumab2 (1.87) Table 1.Demographic and clinicopathological characteristics of patients.Patient CharacteristicsNo. (%) (N = 107)Sex Male83 (77.57)Female24 (22.43)Age(years) <6560 (56.07)≥6547 (43.93)Mean Age64 years oldStage I-III28 (26.17)IV79 (73.83)Histopathology SCLC28 (26.17)LUSC20 (18.69)NS-NSCLC59 (55.14)Treatment Combined with chemotherapy82 (76.64)Combined with targeted therapy30 (28.03)Combined with radiotherapy20 (18.69)Surgery Yes31 (28.97)No76 (71.03)Side effects of immunotherapy Yes36 (33.64)No71 (66.36) Demographic and clinicopathological characteristics of patients. Table 2.The number and proportion of different immunodrugs.ImmunodrugsNo. (%) (N = 107)Tislelizumab20 (18.69)Pembrolizumab22 (20.56)Sintilimab20 (18.69)Camrelizumab22 (20.56)Toripalimab5 (4.67)Durvalumab14 (13.08)Atezolizumab1 (0.93)Nivolumab2 (1.87) The number and proportion of different immunodrugs. Serum ferritin levels Before immunotherapy, patients with lung cancer had a minimum serum ferritin level of 41.80 ng/ml and a maximum of 12,355.04 ng/ml, respectively. The 25% percentile, median, and 75% percentile were 179.20 ng/ml, 301.51 ng/ml and 557.95 ng/ml, respectively. The serum ferritin levels detected at 2–3 treatment cycles (6–9 weeks) were used as the standard during immunotherapy (exception for one case of patient death whose standard was 3 weeks). A ≥ 10% reduction in serum ferritin levels (during immunotherapy vs before immunotherapy) was considered to be downregulated, while a ≥ 10% increase (during immunotherapy vs before immunotherapy) was considered as upregulated. The serum ferritin levels were downregulated in 44 patients and upregulated in 42 patients, leaving 21 cases with insufficient data to infer a conclusion. Correlation between serum ferritin levels and pathological types, tumor stages and PD-L1 expression analysis The patients were divided into 3 groups based on the serum ferritin levels in different pathological types of lung cancer (SCLC, LUSC, NS-NSCLC). The distribution of serum ferritin in the 3 groups was shown using a violin diagram (Figure 1. There were 3 cases where the serum ferritin was more than 2000 ug/ml, and these were defined as 2000 ng/ml in the violin diagram. The results showed no correlation between the pathological types and serum ferritin levels (nonparametric rank sum test, P = .836). Figure 1.a. The number of patients with SCLC, LUSC, and NS-NSCLC was 28, 20, and 59, respectively. NS-NSCLC includes lung adenocarcinoma, large-cell lung cancer, adenosquamous carcinoma lung cancer, and undifferentiated lung cancer. The number of cases of lung adenocarcinoma, large cell lung cancer, lung adenosquamous carcinoma, and undifferentiated lung cancer was 52, 3, 3, and 1, respectively. The distribution of serum ferritin in the three groups was displayed in this violin diagram. The result showed no correlation between different pathological types and serum ferritin levels (nonparametric rank sum test, p = .836). b. The number of patients with stage I-III and stage IV lung cancer was 28 and 79, respectively. The distribution of serum ferritin in different stages of lung cancer (I-III, IV) was shown in this violin diagram. The results showed no significant correlation between the tumor stages and serum ferritin levels, but the result P value was close to 0. 05 (nonparametric rank sum test, P = .053). c. The number of cases of different ranges of PD-L1 expression (PD-L1 < 1%; PD-L1 ≥ 1% and < 50%; PD-L1 ≥ 50%) were 15, 13, and 15, respectively. The distribution of serum ferritin in different PD-L1 ranges was shown in this violin diagram. The result showed no correlation between PD-L1 expression and serum ferritin levels (nonparametric rank sum test, p = .633). d. We collected serum ferritin results from 586 normal people,and the maximum value was 1358.00 ng/ml, and the minimum value was 1.16 ng/ml. The 25% percentile, median, and 75% percentile were 31.06 ng/ml, 89.13 ng/ml and 165.57 ng/ml, respectively. a. The number of patients with SCLC, LUSC, and NS-NSCLC was 28, 20, and 59, respectively. NS-NSCLC includes lung adenocarcinoma, large-cell lung cancer, adenosquamous carcinoma lung cancer, and undifferentiated lung cancer. The number of cases of lung adenocarcinoma, large cell lung cancer, lung adenosquamous carcinoma, and undifferentiated lung cancer was 52, 3, 3, and 1, respectively. The distribution of serum ferritin in the three groups was displayed in this violin diagram. The result showed no correlation between different pathological types and serum ferritin levels (nonparametric rank sum test, p = .836). b. The number of patients with stage I-III and stage IV lung cancer was 28 and 79, respectively. The distribution of serum ferritin in different stages of lung cancer (I-III, IV) was shown in this violin diagram. The results showed no significant correlation between the tumor stages and serum ferritin levels, but the result P value was close to 0. 05 (nonparametric rank sum test, P = .053). c. The number of cases of different ranges of PD-L1 expression (PD-L1 < 1%; PD-L1 ≥ 1% and < 50%; PD-L1 ≥ 50%) were 15, 13, and 15, respectively. The distribution of serum ferritin in different PD-L1 ranges was shown in this violin diagram. The result showed no correlation between PD-L1 expression and serum ferritin levels (nonparametric rank sum test, p = .633). d. We collected serum ferritin results from 586 normal people,and the maximum value was 1358.00 ng/ml, and the minimum value was 1.16 ng/ml. The 25% percentile, median, and 75% percentile were 31.06 ng/ml, 89.13 ng/ml and 165.57 ng/ml, respectively. The patients were divided into 2 groups based on the serum ferritin levels in different stages of lung cancer (I-III, IV). The distribution of serum ferritin in the 2 groups was shown using a violin diagram (Figure 1. There were 3 cases where the serum ferritin was more than 2000 ug/ml, and these were defined as 2000 ng/ml in the violin diagram. The results showed no significant correlation between the tumor stages and serum ferritin levels, but the result P value was close to 0. 05 (nonparametric rank sum test, P = .053). As can be seen from the violin diagram, the distribution of basic serum ferritin level in stage I-III lung cancer patients was lower than that in stage IV lung cancer patients. As we all know, in general, patients with stage IV lung cancer have worse disease treatment response and survival than those with stage I-III lung cancer. Therefore, lower basal serum ferritin levels in lung cancer patients may be associated with better disease treatment response and longer survival, which was consistent with our previous hypothesis. We compiled the cases with PD-L1 expression (43 cases) and divided the patients into 3 groups based on the expression range (PD-L1 < 1%; PD-L1 ≥ 1% and < 50%; PD-L1 ≥ 50%). A violin diagram was used to show the association between serum ferritin and PD-L1 expression (Figure 1. No correlation between PD-L1 expression and serum ferritin level was observed (nonparametric rank sum test, P = .633). Correlation between serum ferritin level and DCR For all patients, serum ferritin levels before immunotherapy were divided into 2 groups with a cutoff of 300 ng/ml.21 The correlation between serum ferritin and DCR was examined using Pearson’s correlation coefficient (Pearson chi-square test, X2 = 4.837, P = .028). The group with serum ferritin <300 ng/ml had a DCR of 84.9% (45/53), while the other group with serum ferritin >300 ng/ml had a DCR of 66.7% (36/54). This indicated that the group with low serum ferritin levels before immunotherapy had better DCR than the group with high serum ferritin levels. We divided 86 patients into two groups based on their serum ferritin upregulation or downregulation during immunotherapy and found similar results (Pearson chi-square test, X2 = 6.475, P = .011). The DCR of the group with downregulated serum ferritin levels was 90.9% (40/44), and that of the other group with upregulated serum ferritin levels was 69.0% (29/42). This showed that the group with downregulated serum ferritin levels during immunotherapy had better DCR than the group with upregulated serum ferritin (Figure 2. These results indicated that low serum ferritin levels before immunotherapy and downregulated serum ferritin levels during immunotherapy correlated with better disease response, which was consistent with our previous hypothesis. To test our hypothesis further, we studied the correlation between serum ferritin levels and patient prognosis. Figure 2.a. Histograms of DCR in all 107 patients separated by serum ferritin levels before immunotherapy. The DCR of the group with serum ferritin <300 ng/ml was 84.9% (45/53), and the DCR of the other group with serum ferritin >300 ng/ml was 66.7% (36/54). This showed that the group with low serum ferritin levels before immunotherapy had better DCR than the group with high serum ferritin levels (Pearson chi-square test, X2 = 4.837, P = .028). b. Histograms of DCR in 86 patients who were divided based on upregulated and downregulated serum ferritin during immunotherapy. The DCR of the group with downregulated serum ferritin levels was 90.9% (40/44), and the other group with upregulated serum ferritin levels was 69.0% (29/42). This suggested that the group with downregulated serum ferritin levels during immunotherapy had better DCR than the group with upregulated serum ferritin (Pearson chi-square test, X2 = 6.475, P = .011). a. Histograms of DCR in all 107 patients separated by serum ferritin levels before immunotherapy. The DCR of the group with serum ferritin <300 ng/ml was 84.9% (45/53), and the DCR of the other group with serum ferritin >300 ng/ml was 66.7% (36/54). This showed that the group with low serum ferritin levels before immunotherapy had better DCR than the group with high serum ferritin levels (Pearson chi-square test, X2 = 4.837, P = .028). b. Histograms of DCR in 86 patients who were divided based on upregulated and downregulated serum ferritin during immunotherapy. The DCR of the group with downregulated serum ferritin levels was 90.9% (40/44), and the other group with upregulated serum ferritin levels was 69.0% (29/42). This suggested that the group with downregulated serum ferritin levels during immunotherapy had better DCR than the group with upregulated serum ferritin (Pearson chi-square test, X2 = 6.475, P = .011). Survival analysis The median PFS for all patients was 6 months. Before immunotherapy, all 107 patients were divided into 2 groups with a serum ferritin level cutoff of 300 ng/ml.21 The results showed that patients with low serum ferritin levels had a longer PFS (median 7 vs 4 months) before immunotherapy than patients with high serum ferritin levels (log-rank test, P = .023). Furthermore, we analyzed the serum ferritin levels before immunotherapy in 79 patients with stage IV disease, which accounted for the majority, with a median of 369.51 ng/ml as the cutoff point. The group with serum ferritin <369.51 ng/ml had a longer PFS (median 6 vs 3.5 months, log-rank test, P = .022). According to the classification standards, 86 patients were divided into 2 groups based on upregulated and downregulated serum ferritin levels during immunotherapy. The results showed that patients with downregulated serum ferritin had a longer PFS (median 9.5 vs 4 months) than patients with upregulated serum ferritin during immunotherapy (log-rank test, P < .001). Subsequently, we performed a survival analysis using the integrated factor of these two prognostic factors. Patients with low serum ferritin levels before immunotherapy and downregulated serum ferritin levels during immunotherapy had the best PFS, while patients with high serum ferritin levels before immunotherapy and upregulated serum ferritin during immunotherapy had the worst PFS (log-rank test, P < .001). The survival curves were shown in Figure 3. Figure 3.Survival curves for PFS. a. Survival analysis of all 107 patients based on serum ferritin levels before immunotherapy, where red indicates serum ferritin levels <300 ng/ml (53 patients) and black denotes serum ferritin levels >300 ng/ml (54 patients). The results showed that patients with lower serum ferritin levels had a longer PFS (median 7 vs 4 months) than patients with higher serum ferritin levels before immunotherapy (log-rank test, P = .023). b. Survival analysis of stage IV patients based on serum ferritin levels before immunotherapy, where red indicates serum ferritin levels <369.51 ng/ml (39 patients) and black denotes serum ferritin levels ≥369.51 ng/ml (40 patients). The group with serum ferritin <369.51 ng/ml had a longer PFS (median 6 vs 3.5 months, log-rank test, P = .022). c. Survival analysis based on upregulated and downregulated serum ferritin during immunotherapy, where red indicates downregulated serum ferritin levels (44 patients) and black denotes upregulated serum ferritin levels (42 patients). The results showed that patients with downregulated serum ferritin had a longer PFS (median 9.5 vs 4 months) than patients with upregulated serum ferritin during immunotherapy (log-rank test, P < .001). d. Survival analysis by integration of serum ferritin levels before and during immunotherapy. Black indicates upregulated serum ferritin levels during immunotherapy and high serum ferritin levels before immunotherapy (17 patients), blue indicates downregulated serum ferritin levels during immunotherapy and low serum ferritin levels before immunotherapy (18 patients), while red indicates downregulated serum ferritin levels during immunotherapy and high serum ferritin levels before immunotherapy or upregulated serum ferritin levels during immunotherapy and low serum ferritin levels before immunotherapy (51 patients). These results showed that patients with low serum ferritin levels before immunotherapy and downregulated serum ferritin during immunotherapy showed the best PFS, while patients with high serum ferritin levels before immunotherapy and upregulated serum ferritin during immunotherapy showed the worst PFS (log-rank test, P < .001). In addition, figure a and figure b patients’ serum ferritin were taken from the values before immunotherapy. Figure c is a comparison of serum ferritin values during 2–3 treatment cycles (6–9 weeks) with pre-immunotherapy serum ferritin values to determine whether patients’ serum ferritin is upregulated or downregulated. Survival curves for PFS. a. Survival analysis of all 107 patients based on serum ferritin levels before immunotherapy, where red indicates serum ferritin levels <300 ng/ml (53 patients) and black denotes serum ferritin levels >300 ng/ml (54 patients). The results showed that patients with lower serum ferritin levels had a longer PFS (median 7 vs 4 months) than patients with higher serum ferritin levels before immunotherapy (log-rank test, P = .023). b. Survival analysis of stage IV patients based on serum ferritin levels before immunotherapy, where red indicates serum ferritin levels <369.51 ng/ml (39 patients) and black denotes serum ferritin levels ≥369.51 ng/ml (40 patients). The group with serum ferritin <369.51 ng/ml had a longer PFS (median 6 vs 3.5 months, log-rank test, P = .022). c. Survival analysis based on upregulated and downregulated serum ferritin during immunotherapy, where red indicates downregulated serum ferritin levels (44 patients) and black denotes upregulated serum ferritin levels (42 patients). The results showed that patients with downregulated serum ferritin had a longer PFS (median 9.5 vs 4 months) than patients with upregulated serum ferritin during immunotherapy (log-rank test, P < .001). d. Survival analysis by integration of serum ferritin levels before and during immunotherapy. Black indicates upregulated serum ferritin levels during immunotherapy and high serum ferritin levels before immunotherapy (17 patients), blue indicates downregulated serum ferritin levels during immunotherapy and low serum ferritin levels before immunotherapy (18 patients), while red indicates downregulated serum ferritin levels during immunotherapy and high serum ferritin levels before immunotherapy or upregulated serum ferritin levels during immunotherapy and low serum ferritin levels before immunotherapy (51 patients). These results showed that patients with low serum ferritin levels before immunotherapy and downregulated serum ferritin during immunotherapy showed the best PFS, while patients with high serum ferritin levels before immunotherapy and upregulated serum ferritin during immunotherapy showed the worst PFS (log-rank test, P < .001). In addition, figure a and figure b patients’ serum ferritin were taken from the values before immunotherapy. Figure c is a comparison of serum ferritin values during 2–3 treatment cycles (6–9 weeks) with pre-immunotherapy serum ferritin values to determine whether patients’ serum ferritin is upregulated or downregulated. Furthermore, we analyzed the serum ferritin levels before immunotherapy in 79 patients with non-small cell lung cancer, which accounted for the majority, with a median of 300.00 ng/ml as the cutoff point.21 The group with serum ferritin <300.00 ng/ml had a longer PFS (median 8 vs 5 months, log-rank test, P = .045). According to the classification standards, 62 patients were divided into 2 groups based on upregulated and downregulated serum ferritin levels during immunotherapy. The results showed that patients with downregulated serum ferritin had a longer PFS (median 12.5 vs 5.5 months) than patients with upregulated serum ferritin during immunotherapy (log-rank test, P < .001). The survival curves were shown in Figure 4. Figure 4.PFS survival curve of non-small cell lung cancer. a. Survival analysis of 79 patients with non-small cell lung cancer based on serum ferritin levels before immunotherapy, where red indicates serum ferritin levels <300 ng/ml (40 patients) and black denotes serum ferritin levels >300 ng/ml (39 patients). The results showed that patients with lower serum ferritin levels had a longer PFS (median 8 vs 5 months) than patients with higher serum ferritin levels before immunotherapy (log-rank test, P = .045). b. Survival analysis based on upregulated and downregulated serum ferritin during immunotherapy, where red indicates downregulated serum ferritin levels (36 patients) and black denotes upregulated serum ferritin levels (26 patients). The results showed that patients with downregulated serum ferritin had a longer PFS (median 12.5 vs 5.5 months) than patients with upregulated serum ferritin during immunotherapy (log-rank test, P < .001). In addition, figure a patients’ serum ferritin were taken from the values before immunotherapy. Figure b is a comparison of serum ferritin values during 2–3 treatment cycles (6–9 weeks) with pre-immunotherapy serum ferritin values to determine whether patients’ serum ferritin is upregulated or downregulated. PFS survival curve of non-small cell lung cancer. a. Survival analysis of 79 patients with non-small cell lung cancer based on serum ferritin levels before immunotherapy, where red indicates serum ferritin levels <300 ng/ml (40 patients) and black denotes serum ferritin levels >300 ng/ml (39 patients). The results showed that patients with lower serum ferritin levels had a longer PFS (median 8 vs 5 months) than patients with higher serum ferritin levels before immunotherapy (log-rank test, P = .045). b. Survival analysis based on upregulated and downregulated serum ferritin during immunotherapy, where red indicates downregulated serum ferritin levels (36 patients) and black denotes upregulated serum ferritin levels (26 patients). The results showed that patients with downregulated serum ferritin had a longer PFS (median 12.5 vs 5.5 months) than patients with upregulated serum ferritin during immunotherapy (log-rank test, P < .001). In addition, figure a patients’ serum ferritin were taken from the values before immunotherapy. Figure b is a comparison of serum ferritin values during 2–3 treatment cycles (6–9 weeks) with pre-immunotherapy serum ferritin values to determine whether patients’ serum ferritin is upregulated or downregulated. Multivariate analysis for all patients showed that serum ferritin level before immunotherapy was an independent prognostic factor (HR = 1.60, P = .041). The stage of lung cancer (HR = 0.53, P = .029) and pathological type were also independent prognostic factors (SCLC: HR = 1, P = .006, LUSC: HR = 0.39, P = .012, NS-NSCLC: HR = 0.42, P = .002). Patient age, gender, treatment strategy, and immune-related side effects were not associated with prognosis. Table 3 contains detailed information.Table 3.Details of the Cox proportional hazard model for 107 patients.variableUnivariate analysisMultivariate analysisHR95%CIPHR95%CIPGender0.740.44–1.250.260 Age (years)0.760.48–1.200.242 Combined with Chemotherapy1.370.79–2.390.267 Combined with targeted therapy0.820.50–1.360.440 Combined with radiotherapy1.340.77–2.330.304 Surgery0.600.35–1.010.054 Side effects of immunotherapy1.030.65–1.650.887 Stage0.560.32–0.950.0330.530.30–0.940.029Serum ferritin1.641.04–2.570.0331.601.02–2.520.041Pathology SCLC1 0.0061 0.006LUSC0.350.17–0.730.0050.390.18–0.810.012NS-NSCLC0.490.29–0.830.0080.420.25–0.730.002The table shows the details of the Cox proportional hazard model for 107 patients. HR and 95% CI were listed. Serum ferritin, lung cancer stage, and pathological type were independent prognostic factors. Bold values indicate P < .05. The grouping of variables in the table is based on Table 1. Serum ferritin: serum ferritin levels before immunotherapy. Abbreviations: HR=hazard ratio; 95% CI = 95% confidence interval. Table 3.Details of the Cox proportional hazard model for 107 patients.variableUnivariate analysisMultivariate analysisHR95%CIPHR95%CIPGender0.740.44–1.250.260 Age (years)0.760.48–1.200.242 Combined with Chemotherapy1.370.79–2.390.267 Combined with targeted therapy0.820.50–1.360.440 Combined with radiotherapy1.340.77–2.330.304 Surgery0.600.35–1.010.054 Side effects of immunotherapy1.030.65–1.650.887 Stage0.560.32–0.950.0330.530.30–0.940.029Serum ferritin1.641.04–2.570.0331.601.02–2.520.041Pathology SCLC1 0.0061 0.006LUSC0.350.17–0.730.0050.390.18–0.810.012NS-NSCLC0.490.29–0.830.0080.420.25–0.730.002The table shows the details of the Cox proportional hazard model for 107 patients. HR and 95% CI were listed. Serum ferritin, lung cancer stage, and pathological type were independent prognostic factors. Bold values indicate P < .05. The grouping of variables in the table is based on Table 1. Serum ferritin: serum ferritin levels before immunotherapy. Abbreviations: HR=hazard ratio; 95% CI = 95% confidence interval. Details of the Cox proportional hazard model for 107 patients. The table shows the details of the Cox proportional hazard model for 107 patients. HR and 95% CI were listed. Serum ferritin, lung cancer stage, and pathological type were independent prognostic factors. Bold values indicate P < .05. The grouping of variables in the table is based on Table 1. Serum ferritin: serum ferritin levels before immunotherapy. Abbreviations: HR=hazard ratio; 95% CI = 95% confidence interval. Similarly, multivariate analysis of the 86 patients’ ferritin change suggested that upregulated or downregulated serum ferritin level during immunotherapy was an independent prognostic factor (HR = 2.65, P = .001). The stage of lung cancer (HR = 0.45, P = .013) and the pathological type (SCLC: HR = 1, P = .025, LUSC: HR = 0.47, P = .080, NS-NSCLC: HR = 0.42, P = .006) were also independent prognostic factors. Table 4 contains detailed information.Table 4.Details of the Cox proportional hazard model for 86 patients.variableUnivariate analysisMultivariate analysisHR95%CIPHR95%CIPStage0.550.31–0.990.0470.450.24–0.850.013Serum ferritin2.811.64–4.82<0.0012.651.50–4.670.001Pathology SCLC1 0.0041 0.025LUSC0.310.14–0.690.0040.470.20–1.090.080NS-NSCLC0.430.24–0.770.0050.420.23–0.780.006The table shows the details of the Cox proportional hazard model for 86 patients. HR and 95% CI were listed. Serum ferritin, lung cancer stage, and pathological type were independent prognostic factors. Bold values indicate P < .05. The grouping of variables in the table is based on Table 1. Serum ferritin: upregulated and downregulated serum ferritin during immunotherapy. Table 4.Details of the Cox proportional hazard model for 86 patients.variableUnivariate analysisMultivariate analysisHR95%CIPHR95%CIPStage0.550.31–0.990.0470.450.24–0.850.013Serum ferritin2.811.64–4.82<0.0012.651.50–4.670.001Pathology SCLC1 0.0041 0.025LUSC0.310.14–0.690.0040.470.20–1.090.080NS-NSCLC0.430.24–0.770.0050.420.23–0.780.006The table shows the details of the Cox proportional hazard model for 86 patients. HR and 95% CI were listed. Serum ferritin, lung cancer stage, and pathological type were independent prognostic factors. Bold values indicate P < .05. The grouping of variables in the table is based on Table 1. Serum ferritin: upregulated and downregulated serum ferritin during immunotherapy. Details of the Cox proportional hazard model for 86 patients. The table shows the details of the Cox proportional hazard model for 86 patients. HR and 95% CI were listed. Serum ferritin, lung cancer stage, and pathological type were independent prognostic factors. Bold values indicate P < .05. The grouping of variables in the table is based on Table 1. Serum ferritin: upregulated and downregulated serum ferritin during immunotherapy. Discussion Serum ferritin, as a biomarker, has a good predictive value for chemotherapy and its other aspects in patients with tumors. Studies have shown that the restoration of normal serum ferritin levels in cancer patients receiving chemotherapy is associated with treatment response.22,23 Ferritin can reduce ROS damage caused by chemotherapeutic drugs and weaken the killing effect on tumor cells.2,24 Furthermore, increased ferritin expression is closely linked to tumor cell proliferation and can reduce the efficacy of chemotherapeutic drugs.25 It is also known that ferritin expression correlates negatively with chemotherapy sensitivity. For example, doxurubicin, an antineoplastic chemotherapeutic drug, works by inducing the expression of P21, a cell cycle regulator, while ferritin inhibits the expression of P21, reducing the drug efficacy. Moreover, the downregulation of ferritin expression via small interfering RNA increased the sensitivity of tumor cells to chemotherapy.26,27 According to research, elevated serum ferritin levels in patients are usually associated with tumor progression and poor survival rates.23,28 Patients with tumors have a poor prognosis when their serum ferritin levels are high before treatment, while those with decreased serum ferritin after treatment have better curative effects. For instance, patients with lung cancer having serum ferritin levels >300 ng/ml, have lower survival rates.21 In patients with hepatocellular carcinoma, the serum ferritin level decreased after effective treatment but increased with tumor progression and recurrence. A persistent increase in serum ferritin was associated with poor prognosis, particularly in patients who were negative for alpha-fetoprotein.29 Furthermore, studies have reported that serum ferritin has a certain significance in the prognostic assessment of patients after surgery. For example, patients with colorectal cancer having high serum ferritin levels after palliative surgery had poor survival times.30 For patients with liver cancer who underwent radiofrequency ablation, serum ferritin can be used as a reference index to assess survival time and recurrence.31 Moreover, serum ferritin level was an important predictor of Trastuzumab efficacy in breast cancer patients with human epidermal growth factor receptor-2 overexpression.28 Based on the above scientific evidence, we demonstrated for the very first time that basal serum ferritin levels before immunotherapy and corresponding changes during immunotherapy are both strong independent prognostic factors for patients with lung cancer and can be used to predict immunotherapy response. Patients with lower basal serum ferritin levels before immunotherapy and those with downregulated serum ferritin during immunotherapy had a longer PFS and a higher DCR. We also demonstrated that considering these two prognostic factors together may be useful to estimate disease survival or recurrence. These findings suggest that serum ferritin may have great clinical value for predicting immunotherapy efficacy in patients with lung cancer. While PD-L1 expression is still the most widely used biomarker for predicting the efficacy of tumor immunotherapy, there are still certain gaps in knowledge. First, some studies found no correlation between elevated PD-L1 expression and higher response rates. For instance, the Checkmate-032 study involving patients with urothelial carcinoma found no significant difference in objective response rates (24.0% vs. 26.2%) between PD-L1 positive (≥1%) and negative subgroups (<1%), respectively.32 Several studies have shown that ICBs can be used to treat lung cancer in whole populations, regardless of PD-L1 expression. Additionally, uneven distribution of PD-L1 (+) tumor or stromal cells may lead to inconsistencies between biopsy specimens and excised tissue.33 Therefore, in cases where a tumor cannot be curatively resected, evaluation of PD-L1 expression may not be reliable.33 Also, it increases the likelihood of false-negative events. Studies have established that PD-L1 expression changes during cancer progression and treatment. For example, Kelly RJ et al. found that concurrent chemoradiotherapy significantly changed the PD-L1 (-) status of half of the patients with advanced esophageal adenocarcinoma cancer to PD-L1 (+) (OR = 6.5, P < .01).34 Furthermore, PD-L1 expression is affected by several other factors. The loss of intracellular oncogenic variants, like the phosphate and tensin homolog deleted on chromosome ten (PTEN) gene and exposure to tumor-infiltrating lymphocytes (TIL)-derived cytokines, both contribute to an increase in PD-L1 expression.35 A few driver mutations may also affect PD-L1 expression. For instance, mutations in the epidermal growth factor receptor (EGFR) can enhance PD-L1 expression and block the activation of TILs.36 Other biomarkers, in addition to PD-L1 expression, have inherent limitations in their use. The KEYNOTE-001 study showed that in PD-L1 non-selected populations, pembrolizumab had a long-lasting clinical benefit in 79% of patients with high TMB. In addition, moderate TMB levels but high ICB response rates were recorded for renal cell carcinoma.37 These findings suggested that PD-L1 and TMB did not concur. More importantly, TMB iterations provide an equal weight to each tumor mutation, but in reality, not all mutations are equally significant in terms of their impact.38 Microsatellite instability testing (MSI) can indicate whether or not PD-1 drugs are working in colorectal cancer, but has little significance for patients with other cancers and lacks broad practicability. In addition to the expression of PD-1/PD-L1 in our samples, few other recognized prognostic markers such as MSI and TMB were detected, so the comparison and correlation study between ferritin and MSI and TMB were not conducted in this study, which is our limitation. As a biomarker, serum ferritin has several advantages over PD-L1, MSI, and TMB tests. First, it is convenient to obtain the results, does not need pathological tissue, can check peripheral blood, and reduce pain in patients. Second, serum ferritin is less expensive and more acceptable. Third, repeated measurement of serum ferritin value is possible. Since this study has only preliminarily confirmed that ferritin is related to the prognosis of lung cancer immunotherapy, we will conduct a prospective clinical study in the next step and compare it with the existing prognostic indicators of cancer immunotherapy to further find out the advantages and disadvantages. Because the mechanism of tumor immunotherapy is very complex, PD-L1 expression alone is inadequate for predicting its efficacy. In this study, we found that the ferritin levels did not correlate with PD-L1 expression but had a good predictive value in the prognosis of lung cancer, suggesting that ferritin could compensate for the lack of PD-L1 expression or that we could combine these two prognostic factors to predict the outcome of immunotherapy in patients with lung cancer. However, we found no correlation between ferritin and pathological types, which may have been due to the small number of cases or tumor species. We, therefore, gathered cases from other tumor species, where the data showed that the prediction of immunotherapy efficacy using ferritin levels was related to different pathological types. According to our findings, the immune prognosis of squamous cell carcinoma was strongly correlated to ferritin levels. However, this correlation may not be the same for adenocarcinoma (data not shown). Additionally, ferritin levels were linked to different immunotherapy outcomes in various tumor species. For example, according to our unpublished data, the prognosis of esophageal cancer immunotherapy was strongly associated with ferritin levels. There may be, however, no strong correlation between immunotherapy prognosis and ferritin levels in gastric cancer. Our future studies will focus on understanding the relationship between ferritin levels and immunotherapy outcomes. This study preliminarily proved that ferritin can predict the prognosis of lung cancer patients. Although we demonstrated that the effect of ferritin on PFS in patients with non-small cell lung cancer and stage IV lung cancer was statistically significant, it was not enough. Our next step is to expand the sample size and further stratify the study of lung cancer patients to refine our research results. In addition, we will continue to publish the relationship between ferritin and prognosis of other cancer types and the relationship between ferritin of different pathological types and prognosis of patients with malignant tumors. In this study, the relationship between ferroptosis and ferritin and immunotherapy is only a scientific hypothesis, and does not involve technical mechanism research, just to see whether there is a correlation and what the trend is. In the future, we will study whether the incidence of ferroptosis can be affected by regulating the level of ferritin through animal specimens in the laboratory, so as to affect the efficacy of immunotherapy. We will also decipher the specific mechanism by which serum ferritin levels predict the prognosis of tumor immunotherapy. According to previous studies, ferritin levels may be elevated in many cancer types,39 including NSCLC.40 However, it is not clear if this is serum ferritin or tumor ferritin. In our study, we used serum ferritin because of its accessibility. Since ferroptosis occurs in tumor tissue, this might have had some impact on our analysis. But, prior research has shown that serum ferritin levels are greatly affected by the tumor tissue. The elevated levels of serum ferritin may be locally released from the tumor site. Patients with malignant tumors having elevated serum ferritin, experienced a ~ 50% drop in ferritin levels after the tumor was surgically removed. This suggests that increased serum ferritin may be associated with tumor masses.41 In addition, approximately 70% of patients with glioblastoma have typically more ferritin in their cerebrospinal fluid (CSF) than in their serum. However, the level of ferritin in CSF is generally low, suggesting that it may be produced and released by local tumors.42 Of course, pathological evidence in tumor tissue is very important, the next step we will be to verify the consistency of ferritin results in tumor tissue and serum in laboratory animal specimens. There are many different and intricate ways that ferritin may affect a patient’s prognosis after immunotherapy. This study hypothesized the relationship between ferroptosis and ferritin and immunotherapy based on previous studies. Therefore, it is very important to analyze the indicators related to ferroptosis in plasma or pathological tissue, but we do not have relevant specimens at present. In the next step, we will verify our hypotheses about ferroptosis and specific regulatory mechanisms in laboratory animal specimens, which will be the focus of our future research. Besides ferroptosis, other important factors are listed as follows: (1) Immune suppression: It has been reported that macrophages in the tumor stroma of breast cancer patients secrete excessive L-ferritin. While the high expression of ferritin prevents the transformation of M2-type macrophages into M1-type, these macrophages exhibit the immunophenotype of M2-type and can play a significant role in cancer.43 (2) Promoting cell proliferation: Ferritin can enter cells through both the canonical transferrin and non-canonical receptor pathways. For instance, Scara5,44 a ferritin receptor that mediates the delivery of non-transferrin iron, may enhance iron storage in cells. It is well established that tumor cells need more iron to participate in metabolism and anabolism, and ferritin transports this iron to promote tumor cell proliferation. (3) Anti-oxidative damage: Studies have shown that ferritin can sequester Fe2+ in solution (catalyzing the conversion of highly toxic Fe2+ to less toxic Fe3+),45 reducing lipid peroxidation caused by Fenton-type reaction.46 Scientists were able to conclusively demonstrate the protective role of ferritin against oxidative damage by exposing endothelial cells to an acute iron load, followed by H2O2 and heme.47 The protective effect of ferritin was also validated in both murine and human leukemia cells.48 (4) Promoting angiogenesis: Ferritin has been found to inhibit the antiangiogenic effects of HKa (a two-chain high-molecular-weight kininogen) in the tumor microenvironment.49 In addition to the above, numerous other avenues need to be further investigated. In conclusion, we found that serum ferritin can be used as a biomarker to predict the efficacy of immunotherapy in lung cancer, and its mechanism needs to be further investigated. Based on our current research results and the prospect of possible future research results, it is possible to add ferritin as a predictive biomarker of immunotherapy effect in the future, or combine it with recognized predictive indicators of immunotherapy effect such as PD-1/PD-L1 expression to predict immunotherapy effect, so as to improve the accuracy of predicting immunotherapy effect. It provides a better reference for us to accept immunotherapy for lung cancer patients in clinical work. In addition, based on our findings, could it be possible in the future to influence ferroptosis by regulating ferritin levels, and thus the efficacy of immunotherapy? If this idea can be realized and applied to the clinic, it will be a boon for cancer patients. Besides, ferritin has several other uses in clinical therapeutics. For instance, it is used as a drug carrier to target tumor cells owing to its unique structural advantages.50 The secrets of ferritin are still worth exploring. Patients and methods Patient information The study included patients who received immunotherapy and were tested for serum ferritin at The First Affiliated Hospital of Shandong First Medical University between April 2018 - August 2022. All patients were diagnosed with lung cancer having pathological or cytological evidence, and had at least one detectable tumor. The tumor stage was classified according to the TNM system defined by the 8th edition of the American Joint Committee on Cancer. The patient’s condition was evaluated using the Response Evaluation Criteria in Solid Tumors (RECIST version 1.1). The disease control rate (DCR) was calculated as the sum of complete response, partial response, and stable disease. Progression-free survival (PFS) was defined as the time from immunotherapy to disease progression or last follow-up. The follow-up data for patients were obtained through medical record reviews and telephone surveys. Measurement of serum ferritin To determine serum ferritin levels, a chemiluminescence particle immunoassay was performed using a human ferritin reagent kit (Abbott Diagnostics, Shanghai, China). Immunohistochemistry was used to detect PD-L1 in the corresponding paraffin-embedded tumor tissues, and expression level of PD-L1 was assessed by experienced pathologists (VENTANA PD-L1 (SP263) Assay, Shanghai, China). Statistical analysis The SPSS 26.0 software (SPSS Inc., Chicago, IL, USA) was used to analyze the data. The median value of serum ferritin levels was defined as the cutoff point, and the patients were divided into two groups.28 The upregulation and downregulation of serum ferritin levels were defined by comparing levels of this biomarker before and during immunotherapy for each patient. The Kaplan-Meier method and log-rank test were performed to obtain the survival curves and analyze the data. The correlation between serum ferritin and DCR was examined using Pearson’s correlation coefficient. The univariate and multivariate analyses for PFS were performed to determine the independent prognostic factor using the Cox proportional Hazards model. The violin diagram and nonparametric rank sum test were used to analyze the relationship between serum ferritin level and PD-L1 expression, and the serum ferritin level distribution in different pathological types and stages of lung cancer. A P-value of < .05 was considered statistically significant. Supplementary Material Ferritin data.docx Click here for additional data file.