Serum lactate dehydrogenase predicts brain metastasis and survival in limited-stage small cell lung cancer patients treated with thoracic radiotherapy and prophylactic cranial irradiation

Small cell lung cancer (SCLC) is characterized by a high risk of brain metastasis and poor survival. This study aims to assess the prognostic role of lactate dehydrogenase (LDH) in limited-stage small cell lung cancer (LS-SCLC) treated with thoracic radiotherapy (TRT) and prophylactic cranial irradiation (PCI). This study retrospectively evaluated 197 consecutive patients who underwent TRT and PCI for LS-SCLC between November 2005 and October 2017. Both pretreatment and maximal serum LDH levels (mLDH) during treatment were checked, and an increased LDH level was defined as more than 240 IU/ml. Clinical factors were tested for associations with intracranial progression-free survival (IPFS) and overall survival (OS) after PCI. The Kaplan–Meier method was used to calculate survival rates, and multivariate Cox regression analyses were carried out to identify variables associated with survival. Of the total patients, 28 had higher pretreatment LDH levels and mLDH levels were increased in 95 patients during treatment. In patients in the normal and elevated mLDH groups, the 1‑, 2‑, and 5‑year IPFS rates were 96.7% vs. 90.1%, 91.7% vs. 73.8%, and 87.8% vs. 61.0% (P < 0.01), respectively. Compared to those with normal LDH levels, patients with increased mLDH levels had a higher cumulative risk of intracranial metastasis (hazard ratio [HR] 3.87; 95% confidence interval [CI] 1.73–8.63; P < 0.01) and worse overall survival (HR 2.59; 95% CI 1.67–4.04; P < 0.01). The factors LDH level at baseline or changes between pretreatment level and maximum level during treatment failed to predict brain metastases or OS with statistical significance. In the multivariate analyses, both mLDH during treatment (HR 3.53; 95% CI 1.57–7.92; P = 0.002) and patient age ≥ 60 (HR 2.46; 95% CI 1.22–4.94; P = 0.012) were independently associated with worse IPFS. Factors significantly associated with worse OS included mLDH during treatment (HR 2.45; 95% CI 1.56–3.86; P < 0.001), IIIB stage (HR 1.75; 95% CI 1.06–2.88; P = 0.029), and conventional radiotherapy applied in TRT (HR 1.66; 95% CI 1.04–2.65; P = 0.034). The mLDH level during treatment predicts brain metastasis and survival in LS-SCLC patients treated with TRT and PCI, which may provide valuable information for identifying patients with poor survival outcomes and possible candidates for treatment intensification.


Introduction
Small cell lung cancer (SCLC) is a highly metastatic and challenging type of carcinoma. While worldwide data for SCLC are not available, it is estimated that SCLC accounts for~15% of lung cancers and causes more than 210,000 deaths per year [1]. The survival outcome for this malignancy is poor, with a 2-year survival rate ranging from 20 to 40% and < 10% for patients with limited-stage disease (LD) and extensive-stage disease (ED), respectively [2,3]. The most important prognostic factors in SCLC are disease stage, performance status (PS) scores, prosoma gastric secretin release peptide (Pro-GRP), neuron-specific enolase (NSE), and lactate dehydrogenase (LDH) levels [4][5][6].
Brain metastases (BMs) are common in SCLC, with 10% of patients presenting with this aggravation at the time of diagnosis and an additional 40-50% subsequently developing it [7,8]. Prophylactic cranial irradiation (PCI) is also part of the standard management in most patients with non-metastatic SCLC who respond to initial treatment, as it significantly reduces the risk of BMs and improves survival [9,10]. Although PCI is considered an effective method, some related studies have pointed out that the incidence of BMs even after PCI was still 4% at 1 year, 30% at 2 years, 11.2-38% at 3 years, and 44% at 4 years [11][12][13][14]. Therefore, it is meaningful to find a prognostic factor to predict BMs in these patients.
A study conducted by Forkasiewicz et al. found that lactate dehydrogenase (LDH), which regulates the processing of glucose to lactic acid, is commonly increased in cancer patients and correlated with poor clinical outcomes and resistance to therapy [15]. A recent study has indicated that LDH was a powerful predictor for overall survival (OS) after whole-brain radiation therapy (WBRT) in SCLC patients with BMs [16]. Therefore, we hypothesize that there is a clinical connection between the level of serum LDH and BM in patients diagnosed with SCLC.
Based on the aforementioned background data regarding this disease, the purpose of the present study was to identify LDH as a potential factor predicting BM and overall survival (OS) of LS-SCLC after thoracic radiotherapy (TRT) and PCI.

Materials and methods
Between November 2005 and October 2017, we identified 207 consecutive SCLC patients who underwent TRT and PCI in Zhejiang Cancer Hospital. All patients had signed informed consent for TRT and PCI. Fig. 1 is the CONSORT diagram for patient selection. We excluded 6 patients who did not have serum LDH tests before or during treatment and 4 patients who had not completed PCI or TRT for various reasons. Therefore, a total of 197 patients were eligible for this analysis. This study has been approved by the Ethics Committee of Zhejiang Cancer Hospital and designed according to the principles of the Declaration of Helsinki. The individuals involved also signed informed consent for radiotherapy and chemotherapy.
All patients were pathologically diagnosed with LS-SCLC without BMs based on computed tomography (CT) and/or magnetic resonance imaging (MRI) findings. For pretreatment TMN staging, brain MRI (preferred) or brain CT with contrast, thoracic CT with contrast, whole-abdomen CT with contrast (preferred) or whole abdominal ultrasound, cervical lymph node ultrasound, positron-emission tomography (PET; preferred), or emission computed tomography (ECT) were all required. Serum LDH test data before treatment and during treatment were available. The upper limit of normal value (ULN) for LDH is 240 IU/L and the maximal serum LDH level (mLDH) was defined as the maximal LDH level tested from the beginning of radiotherapy or chemotherapy to the end of treatment.
Additionally, IPFS was defined as the interval from pathological diagnosis to the onset of brain metastases or death or the last follow-up date. Diagnosis of intracranial progression mainly depends on imaging, but when the BM symptoms were identified before the imaging diagnosis, the first day of symptoms was considered the cutoff point. Within 1 year of the end of treatment, the patient underwent brain MRI (preferred) or brain CT every 3-4 months, and from the second year onwards, brain MRI (preferred) or brain CT was performed every 6 months.  Furthermore, PET was not commonly used during follow-up unless metastases were detected by other tests and the patient's financial conditions allowed. Other items (e.g., enhanced CT of chest and abdomen) were performed every 3-4 months within 2 years of the end of treatment and every 6 months after 2 years. Most of our patients had good follow-up compliance, and very few patients did not participate in the outpatient clinic for follow-up. Serum LDH levels at baseline and before each treatment (each cycle of chemotherapy and radiotherapy) were routinely measured as a part of biochemical tests using a Hitachi Modular 7600 Chemistry Analyzer. Therefore, the frequency of test- ing was regular, each 3-4 weeks, the same frequency as chemotherapy.
On the occasion of abnormal biochemical indexes, such as liver and kidney function, biochemical tests were repeated after symptomatic therapy. In general, patients with LS-SCLC usually require at least one cycle of neoadjuvant chemotherapy, followed by thoracic radiotherapy with or without concurrent chemotherapy. A total of six cycles of chemotherapy (including concurrent chemotherapy) were required so that PCI could finally be performed. The implementation time of concurrent chemoradiotherapy can be after any cycle of chemotherapy. In total, 97 patients did not complete the six cycles of chemotherapy, 17 patients underwent PCI before chemotherapy was completed, and none of the patients underwent concurrent chemotherapy during PCI.
All patients were typically treated with PCI at a photon energy of 6 MV and laterally opposed treatment fields that encompassed the entire brain. The prescribed dose was calculated at the isocenter of the radiation fields based on daily treatments. While 7 (3.6%), 80 (40.6%), and 15 (7.6%) pa-tients in the normal group underwent PCI with a lower standard dose (SD; 20 Gy/10 fractions or 24 Gy/12 fractions), medium SD (25 Gy/10 fractions or 30 Gy/10-15 fractions) and higher SD (36 Gy/18 fractions or 40 Gy/20 fractions), respectively, 3 (1.5%), 79 (40.1%), and 13 (6.6%) patients in the elevated group underwent PCI with a lower SD, medium SD, and higher SD, respectively. The median biologically effective dose (BED) was 36 (range 24-48) Gy for both groups when prescription doses were corrected to the BED using the linear quadratic model with an assumed α/β ratio of 10 Gy for tumor tissue.

Statistical analysis
The data used in this study are reported as median (range) or number (percentage). Time-to-event analyses were performed from the start of TRT to the emergence of the event. Descriptive statistical analyses were applied to characterize the patients in the normal and elevated groups. Chisquared test, which was carried out with SPSS 22.0 software (IBM Corporation, Armonk, NY, USA), was adopted for estimation of the differences in clinical characteristics (smoking index, ECOG-PS, TMN stage, number of cycles of chemotherapy, chemotherapy regimen, combined modality of Chemo-RT, PCI dose, and demographic variables). The Kaplan-Meier method and the log-rank test were used to compare the curves for intracranial progression-free survival (IPFS) and OS. Thereafter, potential prognostic factors were evaluated using the Cox proportional hazards model, and the results were reported as hazard ratios (HRs) and the corresponding 95% confidence intervals (CI). Significant factors identified in univariate analyses were included in the multivariate model. GraphPad Prism 7, launched by GraphPad Software (San Diego, CA, USA), was used to draw the forest figure of survival analysis. We considered the differences statistically significant if P-values < 0.05.

Patient characteristics and outcomes
The median number of LDH tests was 7 (range 4-12) for all patients during treatment, including those in the elevated and normal groups. The mLDH level during treatment (median 233 IU/L, range 74-2327) was significantly higher than the pretreatment LDH level (median 183 IU/L, range 73-1999), and the average level of LDH during treatment (median 172 IU/L, range 89-984) decreased slightly compared to the pretreatment level (median 183 IU/L, range 73-1999). The levels of LDH during treatment might be associated with therapeutic effect, glycolytic activity, liver function, body inflammation, tumor burden, and necrosis.
While 28 patients presented higher pretreatment LDH levels (≥ 1 ULN), serum mLDH levels were promoted in 95 patients during treatment. Both pretreatment LDH and mLDH during treatment were ≥ 1ULN in 15 patients and < 1ULN in 88 patients. Pretreatment LDH ≥ 1ULN but mLDH during treatment < 1ULN was identified in 13 patients. Also, pretreatment LDH < 1ULN but mLDH during treatment ≥ 1ULN was identified in 80 patients. As shown in Table 1

mLDH during treatment is associated with a higher risk of brain metastasis and predicts IPFS
The upper limit of the normal range was chosen as the cutoff value for LDH based on the results of evaluation of various cutoff values. As shown in Table 2, univariate analyses revealed that longer IPFS was associated with mLDH during treatment < ULN (P < 0.01) and age < 60 years (P < 0.01). In patients in the normal and elevated LDH groups, the 1-, 2-, and 5-year IPFS rates were 96.7% vs. 90.1%, 91.7% vs. 73.8%, and 87.8% vs. 61.0% (P < 0.01), respectively (as shown in Fig. 3a). Compared to those patients with normal LDH levels, patients with increased mLDH levels had a higher cumulative risk of intracranial metastasis (hazard ratio [HR] 3.87; 95% confidence interval [CI] 1.73-8.63; P < 0.01). No significant impact on IPFS after TRT and PCI was observed for pretreatment LDH level or changes between pretreatment LDH and maximum LDH during treatment level (as shown in Table 2).
The results of multivariate analyses are shown in Fig. 3b. As shown, mLDH level during treatment ≥ ULN (P = 0.002) and age ≥ 60 (P = 0.012) were identified as significant independent predictors of poor IPFS.

Univariate and multivariate models for overall survival
As shown in Fig. 4a,b, mLDH levels during treatment were associated with worse survival. In patients in the normal and elevated mLDH groups, the 1-, 2-, and 5-year OS rates were 89.6% vs. 79.8%, 74.2% vs. 51.1%, and 58.2% vs. 29.4% (P < 0.01), respectively. However, no sig-nificant impact on OS after TRT and PCI was observed for pretreatment LDH levels or changes between pretreatment LDH and maximum LDH levels during treatment ( Table 2). Compared to patients with normal LDH levels, patients with increased mLDH levels had a higher cumulative risk of death (HR 2.59; 95% CI 1.67-4.04; P < 0.01). Factors associated with improved OS were mLDH during treatment < ULN (P < 0.01), IA-IIIA stage at initial diagnosis (P = 0.02), and three-dimensional conformal or IMRT applied in TRT (P = 0.02). Other factors, such as age and chemotherapy regimens, were suspected predictors, although the P-value was slightly greater than 0.05 ( Table 2). Fig. 4c represents the OS results of multivariate analysis. In addition to the statistically significant indicators in univariate analysis, other factors that might affect the results, such as age, sex, chemotherapy regimen, and pretreatment LDH, were also included in this analysis. As shown in Fig. 4c, mLDHs level during treatment ≥ ULN (P = 0.002), conventional technique applied in TRT (P = 0.034), and IIIB stage at initial diagnosis after treatment (P = 0.029) were identified as significant independent predictors of poor OS.

mLDH during treatment is associated with a higher risk of extracranial metastasis and predicts extracranial progression-free survival
In patients in the normal and elevated mLDH groups, the 1-, 2-, and 3-year extracranial progression-free survival (ECPFS) rates were 71.6% vs. 50.9%, 63.6% vs. 36.3%, and 60.6% vs. 31.8% (P < 0.01), respectively (as shown in Supplementary Figure A). Compared to those patients with normal LDH levels, patients with increased mLDH levels had a higher cumulative risk of extracranial metastasis (HR 2.34; 95% CI 1.57-3.49; P < 0.01). No significant impact  Figure C).

Discussion
As a key enzyme in glycolysis, LDH has been reported to be enhanced in transformed cells and play a vital role in tumor initiation, proliferation, invasion, and metastasis [17]. Furthermore, serum LDH has been proven to be a powerful -BMs brain metastases, EP etoposide and platinum, TRT thoracic radiotherapy, IMRT intensity-modulated radiotherapy, BED biological effective dose, LDH lactate dehydrogenase, ULN upper limit of normal value predictor in various cancers, and some studies have also confirmed that serum LDH could strongly predict survival in LS-SCLC [18][19][20][21][22][23]. However, none of these studies have identified LDH as a prognostic indicator to predict brain metastasis and survival in LS-SCLC after PCI or explored the relationship between LDH and brain metastasis. Elevated LDH levels represent higher glycolysis activity, which might promote cancer invasion and metastasis. Different scholars have looked into this phenomenon and concluded that energy metabolism plays an important role in cerebral metastasis [24,25]. In such cases, glycolysis inhibition might be a useful strategy to reduce the risk of cerebral metastasis in LS-SCLC. Our previous study [26] revealed that oxamate, an inhibitor of LDH-A, significantly suppressed the proliferation of NSCLC cells while it exerted much lower toxicity in normal cells. LDH-A inhibition resulted in ATP reduction and reactive oxygen species (ROS) burst in cancer cells, which led to apoptosis and G2/M arrest and increased radiosensitivity in NSCLC cells [27].
Up to now, scholars and clinicians have still not found the best method to select early-stage SCLC patients with good prognoses for possible avoidance of PCI. Previous metaanalysis identified five retrospective studies and included a total of 1691 patients, among which 315 of received PCI. For all the resected patients, PCI was associated with improved overall survival (HR 0.52, 95% CI 0.33-0.82) and reduced brain metastasis risk (RR 0.50, 95% CI 0.32-0.78). However, regarding p-stage I patients, no survival benefit was brought by PCI (HR 0.87, 95% CI 0.34-2.24) [28]. Due to this study, NCCN guidelines 2019 version 1 did not recommend PCI for p-stage I (T1-2N0M0) patients who had undergone radical surgical interventions (category IIA). The present study showed that elevated mLDH levels during treatment might indicate disease recurrence and brain metastasis. Therefore, for those patients, MRI is necessary. Recently, Anami et al. evaluated 48 consecutive patients who underwent WBRT for BMs from SCLC, and the results revealed that the presence of symptoms due to BMs and LDH values independently predicted prognosis [16]. Suzuki et al. also identified that high pretreatment platelet counts (1.649, 95% CI 1.130-2.408; P = 0.010) and pretreatment LDH > 543 U/L (HR 1.870, 95% CI 1.290-2.710; P = 0.001) were associated with increased rates of brain metastasis in patients with SCLC with no evidence of brain disease at diagnosis [29]. These studies suggested some clinical links between BM and elevated LDH. In our research, elevated mLDH levels during treatment were treated as a significant independent prognostic indicator for IPFS in LS-SCLC after TRT and PCI (HR for IPFS 3.53, 95% CI 1.57-7.92; P = 0.002). These data further confirm the connections between BM and elevated LDH. Thus, for the patients with elevated LDH during treatment, more positive therapies should be administered to reduce the risk of BM. At least PCI, which has been proven to improve IPFS, should been applied urgently. Other options, such as LDH inhibitors or glycolysis inhibitors, can be used for BM prevention. At present, these drugs are still not widely available, and there is a lack of research focusing on these substances, randomized trials on the use of relevant drugs for BM prevention should be conducted in the future.
In this study, the pretreatment LDH level or changes between pretreatment and maximum LDH level during treatment predicted IPFS and OS without statistical significance. Our outcomes are not consistent with the results reported by Sagman et al. and He, et al. [21,22]. In Sagman's study, patients with LS-SCLC and elevated levels of pretreatment LDH manifested a higher relative death rate (1.63:1) when compared to patients with LS-SCLC and LDH in the normal range (P = 0.0083), but the survival of patients with extensive-stage disease did not differ between patients with normal and elevated levels of LDH (P = 0.273). Contrastingly, in He et al.'s study, multivariate analysis revealed that pretreatment LDH ≥ 215.70 U/L was an independent prognostic factor for poor survival (HR 1.468, 95% CI 1.069-2.017; P = 0.018). The subgroup analysis showed that pretreatment LDH level was significant for predicting survival in both limited and extensive disease. Further, Suzuki et al. [29] also identified pretreatment LDH as an influential prognostic factor for BM in patients with SCLC with no evidence of brain disease at diagnosis. Our results are different from others reported in the aforementioned studies, probably because in these papers, the sample of patients with SCLC included all TMN or limited stages without PCI, and diverse samples may make a difference in the prediction of BM and survival. Other factors, such as small patient samples and inconsistency of clinicopathological parameters, may also contribute to the different results.
In addition, our study showed that the 1-, 2-, 3-, and 5-year IPFS rates were 94.0%, 83.5%, 79.5%, and 75.2%, and the 1-, 2-, and 5-year OS were 84.7%, 62.6%, and 46.5%, respectively. The 2-year OS rate of our study was relatively high, which was much better than the one reported by Kamran et al. [30] (62.6% vs. 47%). We assume that there may be several possible reasons behind this discrepancy, including the fact that 40% of the patients included in their study did not undergo PCI, while all of the patients in our study completed PCI. This difference may have directly affected our results, as we know, PCI can improve OS of LS-SCLC by 5.4% [31]. Secondly, 18% of Kamran et al.'s patients have an ECOG-PS of 2-3, while only 5.1% of patients in our study have ECOG-PS of 2, and since ECOG-PS is also a prognostic factor, these different rates may have also impacted the final results. Lastly, the proportion of stage IA-IIIA patients in our study is much higher than that in Kamran et al.'s study (40% vs. 17.3%), and TMN stage is considered a very powerful predictor by many scholars, which is in agreements with the preconditions applied in our analyses. In conclusion, this retrospective dataset provides evidence that maximum elevated LDH levels during treatment of patients with LD SCLC may predict for the development of brain metastases and survival. However, given the limited number of patients and unseasonal follow-up of very few patients, our findings still need to be confirmed by more studies. In addition, future research should develop a comprehensive scoring tool to assist clinicians to decide whether to administrate PCI in LS-SCLC patients.