Serum lactate dehydrogenase and vascular endothelial growth factor A level as a predictor of treatment response in locally advanced-stage World Health Organisation type 3 nasopharyngeal carcinoma

Abstract

Introduction

This study aims to evaluate the association of baseline serum vascular endothelial growth factor A (VEGF-A) and lactate dehydrogenase (LDH) levels with subsequent response to sequential neoadjuvant chemotherapy followed by radiotherapy in patients with locally advanced-stage nasopharyngeal carcinoma (LA-NPC).

Material and methods

A prospective nested case-control study using consecutive sampling with an outcome-based quota cap was conducted involving patients with World Health Organisation type 3 NPC, stage III–IVA. Patients were treated with three cycles of taxane-cisplatin-based neoadjuvant chemotherapy followed sequentially by definitive intensity-modulated radiotherapy. Serum VEGF-A and total LDH levels were measured prior to therapy using enzyme-linked immunosorbent assay. Treatment response was assessed using Response Evaluation Criteria in Solid Tumours 1.1 criteria at a fixed time point of 8 weeks after the completion of all planned treatment, and categorised as positive (complete or partial response) or negative (stable or progressive disease). Statistical analyses including bivariate testing, receiver operating characteristic analysis, and multivariate logistic regression.

Results

A total of 58 patients were analysed. In univariate binary logistic regression assessments, high exploratory baseline serum VEGF-A (≥ 128.525 ng/l; p = 0.011) and LDH (≥ 66.485 U/l; p = 0.009) levels were significantly associated with subsequent negative treatment response. In a pre-specified multivariable model adjusted for age and VEGF-A, an elevated baseline total LDH level remained a significant independent predictor of treatment non-responsiveness (adjusted OR = 4.68; 95% CI: 1.46–15.00; p = 0.009), whereas VEGF-A lost statistical significance (p = 0.257).

Conclusions

Elevated baseline serum VEGF-A and LDH protein levels are associated with non-responsiveness to sequential neoadjuvant chemotherapy and radiotherapy in LA-NPC. Pre-treatment total LDH acts as a significant, independent exploratory indicator of treatment non-responsiveness.

Introduction

Nasopharyngeal carcinoma (NPC) is a malignant epithelial tumour originating from the nasopharyngeal mucosa and represents a distinct entity among head and neck cancers due to its unique epidemiology, aetiology, and biological behaviour. Nasopharyngeal carcinoma shows marked geographic and ethnic predilection, with high incidence rates reported in Southern China, Southeast Asia, North Africa, and the Arctic region [1]. Indonesia is one of the endemic regions, where NPC constitutes the most common malignancy of the ear, nose, and throat region and ranks among the top five cancers overall [2].

Despite advances in diagnostic imaging and radiotherapy techniques, the majority of NPC patients in endemic countries present at an advanced stage (stage III–IV) at initial diagnosis, primarily due to nonspecific early symptoms and the anatomically concealed location of the nasopharynx [3]. Advanced- stage disease is associated with poorer prognosis and higher rates of locoregional failure and distant metastasis compared with early-stage NPC [4].

Radiotherapy remains the cornerstone of NPC treatment because of the tumour’s inherent radiosensitivity. However, for advanced-stage tumours/NPC, radiotherapy alone yields suboptimal outcomes, prompting the integration of chemotherapy into treatment protocols [5]. Neoadjuvant or concurrent chemoradiotherapy has become the standard of care for locally advanced-stage NPC (LA-NPC), improving overall survival and disease control compared with radiotherapy alone [6]. According to Dai et al. [7], the use of induction chemotherapy followed by radiotherapy and concomitant chemoradiation have similar effectiveness on LA-NPC. On the other hand, the response to treatment among patients may vary. Susanto et al. [8] in their report found in LA-NPC that out of 383 patients, 82% achieved complete response, 13.1% achieved partial response, and 2.9% and 2.1% had progressive disease and stable disease consecutively.

The heterogeneity in treatment response highlights the need for reliable biomarkers that can predict therapeutic outcomes and guide individualised treatment strategies. Tumour angiogenesis and altered cancer metabolism are two hallmarks of cancer progression and treatment resistance [9]. Vascular endothelial growth factor A (VEGF-A) is a key mediator of angiogenesis and plays a central role in tumour growth, invasion, and metastasis by promoting neovascularization and increasing vascular permeability [10]. Overexpression of VEGF-A has been associated with a poor prognosis in various solid tumours, including NPC [11].

Lactate dehydrogenase (LDH) is a cytoplasmic enzyme involved in anaerobic glycolysis, catalysing the conversion of pyruvate to lactate. Elevated LDH levels reflect increased glycolytic activity, tumour hypoxia, and aggressive tumour biology, all of which contribute to radioresistance and chemoresistance [12]. Serum LDH has been proposed as a simple, inexpensive prognostic biomarker in several malignancies, including lymphoma, lung cancer, and NPC [12, 13].

World Health Organisation (WHO) type 3 NPC (undifferentiated carcinoma) is the predominant histological subtype in endemic regions, including Indonesia, and is strongly associated with Epstein-Barr virus (EBV) infection [2, 14]. Although WHO type 3 is generally considered more radiosensitive than keratinizing subtypes, treatment failure and disease progression remains common in advanced stages [15, 16]. Identifying biomarkers that predict response specifically in WHO type 3 LA-NPC is therefore of particular clinical relevance.

This study aims to evaluate high serum VEGF-A and LDH levels as risk factors for poor response to neoadjuvant chemoradiotherapy in patients with WHO type 3 LA-NPC..

Material and methods

Study design and setting

This prospective nested case-control study was conducted at RSUDP NTB in Indonesia, between August 2024 and August 2025.

Study population

The target population consisted of patients diagnosed with LA-NPC. Eligible participants were aged ≥ 18 years with histopathologically confirmed WHO type 3 NPC and advanced clinical stage (stage III–IVA) based on the American Joint Committee on Cancer 8^th edition staging. To determine the sample size, baseline parameters were derived from the epidemiological reports by Susanto et al. [8], where the cumulative proportion of positive therapeutic response (complete response [CR] + partial response [PR]) was 95.1% (P₁), compared to 4.9% (P₂) for non-responsive disease (stable disease [SD] + progressive disease [PD]). Using Lemeshow’s equation with a type I error rate (α) of 5% and a power of 90%, the formal calculation generated a minimal sample size of 4 participants per cohort. Recognising that a sample of this size is statistically underpowered for robust laboratory comparisons, the target sample size was pragmatically adjusted upward to 29 participants per cohort to satisfy the distribution assumptions of subsequent biochemical comparative analyses and to mitigate potential outlier-induced bias. This inflation ensures sufficient statistical power to satisfy the distribution assumptions of subsequent comparative analyses for serum LDH and VEGF markers, while mitigating potential confounding from extreme outlier values.

To achieve this balanced allocation, a consecutive sampling strategy with an outcome-based quota cap was implemented. Patients meeting the clinical criteria were screened prospectively and consecutively at the immediate completion of their treatment regimen. Upon evaluation of their therapeutic response, subjects were assigned to either the positive or negative response group. Enrolment for the positive response cohort was closed immediately upon reaching its targeted quota of 29 patients, whereas consecutive screening and recruitment continued exclusively for the rarer negative response cohort until its respective quota of 29 patients was fully satisfied.

Patients were excluded if they had a history of other malignancies, distant metastasis at diagnosis (stage IVB), prior chemotherapy or radiotherapy, severe comorbid conditions precluding chemoradiotherapy, or incomplete clinical or laboratory data. Furthermore, to ensure a highly homo- genous cohort and eliminate the confounding effects of treatment protraction, we excluded patients who experienced severe toxicities requiring chemotherapy dose reductions, treatment delays, or radiotherapy interruptions.

Treatment protocol

All enrolled patients received standardised neoadjuvant (induction) chemotherapy followed sequentially by definitive radiotherapy alone. No concurrent and adjuvant chemo-therapy was administered. The neoadjuvant chemo- therapy phase consisted of three cycles of a taxane- cisplatin-based regimen administered every three weeks. The regimen comprised paclitaxel in combination with either cisplatin or carboplatin, with dosages determined by body surface area or area under the curve (AUC), in accordance with established institutional protocols. Following the completion of the three chemotherapy cycles, radiotherapy was delivered using intensity-modulated radiation therapy. A total definitive prescription dose of 70 Gy was delivered to the primary tumour and gross nodal volumes in 33 fractions over approximately 6.5 weeks.

Biomarker measurement

Venous blood samples were collected from all consecutively screened patients at the time of initial diagnosis, prior to neoadjuvant chemoradiotherapy. To obtain serum, blood samples were allowed to clot at room temperature for 20 min, followed by centrifugation at 3,000 rpm for 15 min. The separated serum supernatants were immediately aliquoted into sterile cryotubes and stored in a deep freezer at –80°C to prevent protein degradation.

Following the definitive clinical and radiological response evaluations at 8 weeks post-radiotherapy, the final comparative groups were locked. Baseline serum samples corresponding strictly to the selected 58 patients (29 objective responders and 29 patients with stable disease) were subsequently retrieved from storage for analysis. Quantitative determination of baseline total human serum LDH protein concentration and serum VEGF-A levels was performed using commercially available sandwich ELISA kits from Bioassay Technology Laboratory (BT Lab; Shanghai, China), strictly following the manufacturer’s protocols. Total LDH was quantified using the LDH ELISA Kit (Catalogue No. E0183Hu), which operates on a colorimetric readout at an absorbance of 450 nm, with an analytical detection range of 2–600 U/l.

Outcome assessment

Therapeutic response was assessed using the Response Evaluation Criteria in Solid Tumours version 1.1, based on clinical examination and radiological imaging Eisenhauer et al., 2009 [17]. This definitive assessment was performed at a fixed time point of 8 weeks (2 months) after the completion of all planned treatments. Responses were classified as CR, PR, SD, or PD. For analysis, CR and PR were categorised as positive response, while SD and PD were categorised as negative response. Only patients who successfully completed the entire sequence of three chemotherapy cycles and all 33 fractions of radiotherapy without treatment-interrupting toxicities progressed to the final outcome assessment.

Objective tumour response was determined using contrast-enhanced computed tomography scans of the head and neck, supplemented by flexible fiberoptic nasopharyngoscopy. To minimise bias, all radiological images were independently reviewed and graded by a senior independent radiologist who was completely blinded to the patients’ laboratory biomarker results (serum LDH and VEGF levels) and clinical group allocations.

Statistical analysis

Statistical analyses were performed using SPSS version 30. Continuous variables (age, VEGF-A level, and LDH level) were tested for normality using the Kolmogorov-Smirnov test. Continues variables with non-normal distributions were presented as medians and range, and comparative analyses between groups were performed using the non-parametric Mann-Whitney U test.

Categorical variables (sex, clinical stage, symptom duration, and dichotomised biomarker classifications) were expressed as frequencies and percentages. Differences in categorical distributions between the positive and negative response cohorts were evaluated using the χ² test or Fisher’s exact test where appropriate; no normality transformations or tests were applied to categorical structures. Receiver operating characteristic curve analysis using Youden’s J statistic was deployed to identify exploratory biomarker cut-off thresholds. Finally, a pre-specified multivariable binary logistic regression model was constructed to determine independent predictors of therapeutic failure. Statistical significance for all tests was set a priori at p < 0.05.

Results

Patient characteristics

A total of 58 patients with WHO type 3 LA-NPC were included in the analysis. Reflecting the outcome-based quota sampling strategy, the distribution of therapeutic response was perfectly balanced between two comparative cohorts. Twenty-nine patients (50.0%) demonstrated a positive objective response, consisting of 23 patients (79.3%) with a CR and 6 patients (20.7%) with a PR. Conversely, the remaining 29 patients (50.0%) demonstrated a non-responsive pattern consisting exclusively of SD. No patients presenting with early PD were evaluated in this cohort. Most patients were male (70.7%) and aged ≥ 45 years (60.4%). The characteristics of subjects in this study were shown in detail in Table 1.

Table 1.

Characteristics of patients

Characteristics		n	Percentage	Statistical value/normality (p)
Age
	< 45	23	39.6	–
	≥ 45	35	60.4
	Mean	–	–	46.2 ±13.1 years
Sex
	Male	41	70.7	–
	Female	17	29.3
Stage
	III	26	44.8	–
	IV A	32	55.2
Time when the first symptoms appear
	< 6 months	23	39.6	–
	≥ 6 months	35	60.4
VEGF-A level
	Median (IQR)	–	–	126.9 (101.9–204.3)/p < 0.001
LDH level
	Median (IQR)	–	–	69.0 (47.7–131.8)/p < 0.001

IQR – interquartile range, LDH – lactate dehydrogenase, VEGF – vascular endothelial growth factor

VEGF-A serum level

The baseline distribution of total serum VEGF-A across the entire cohort was non-normal (p < 0.001), with an overall mean of 173 ng/l and a median of 127 ng/l. Pre-treatment baseline serum VEGF-A levels were significantly higher in the group that subsequently demonstrated stable disease at 8 weeks (median: 138.46 ng/l; range: 22.78–644.94 ng/l) than in the group that achieved a positive response (median: 114.67 ng/l; range: 18.28–287.49 ng/l; p = 0.011) (Figure 1).

Figure 1.

Box plot of lactate dehydrogenase and vascular endothelial growth factor A serum level, stratified by the response group

The baseline serum VEGF-A level exhibited a modest predictive performance for treatment failure, with an AUC of 0.694 (95% CI: 0.560–0.829; p = 0.011). The optimal exploratory cut-off value derived via Youden’s J statistic was 128.5 ng/l. This threshold provided a sensitivity of 62.1% (95% CI: 42.3–79.3) and a specificity of 69.0% (95% CI: 49.2–84.7). The corresponding positive predictive value (PPV) was 66.7%, and the negative predictive value (NPV) was 64.5%. Consistently, these findings suggest that baseline VEGF-A may serve as a preliminary, exploratory indicator of subsequent stable disease rather than an independent tool for personalised treatment stratification (Figure 2).

Figure 2.

Receiver operating characteristic of vascular endothelial growth factor A level in patients with locally advanced-stage nasopharyngeal carcinoma

Total lactate dehydrogenase serum level

The baseline distribution of total serum LDH across the entire cohort was non-normal (p = 0.001), with a mean of 118 U/l and overall median level of 69 U/l. According to the Mann-Whitney U test, the total serum LDH protein level was significantly elevated in the negative response cohort, demonstrating a median level of 78.62 U/l (range: 23.20–543.42 U/l), compared to 53.40 U/l (range: 17.80–195.05 U/l) in positive response cohort (p = 0.009) (Figure 3). This indicates that higher pre-treatment total LDH protein concentrations are significantly associated with therapeutic failure.

Figure 3.

Receiver operating characteristic for lactate dehydrogenase in locally advanced-stage nasopharyngeal carcinoma

The receiver operating characteristic analysis demonstrated a modest discriminative capacity for baseline total serum LDH protein levels in predicting subsequent therapeutic failure (p = 0.009). The analysis yielded an AUC of 0.699 (95% CI: 0.560–0.837). Utilising Youden’s J statistic, the optimal exploratory cut-off point was determined to be 66.5 U/l. At this specific threshold, baseline total LDH achieved a sensitivity of 72.4% (95% CI: 52.8–87.3) and a specificity of 62.1% (95% CI: 42.3–79.3). Within this study’s cohort, the PPV was 65.6% and the NPV was 69.2%. These metrics indicate that while elevated pre-treatment total LDH holds a significant association with treatment non-responsiveness, its singular capacity for definitive clinical prediction remains limited.

Association of variables with treatment response

Bivariate analysis demonstrated that male sex (OR = 3.39; p = 0.043), high VEGF-A (OR = 3.64; p = 0.014), and high LDH levels (OR = 4.30; p = 0.008) were significantly associated with negative response to neoadjuvant chemoradiotherapy. The remaining variables showed no significant association with response to therapy (Table 2).

Table 2.

Response to therapy based on several variables

Parameters		Negative response		Positive response		p (χ2)
		n	Percentage	n	Percentage
Age
	< 45	15	25.8	8	13.8	0.060
	≥ 45	14	24.1	21	36.2
Sex
	Male	24	41.4	17	29.3	0.043
	Female	5	8.6	12	20.7
Stage
	III	8	13.8	14	31.1	0.104
	IV A	21	29.3	15	25.9
Time when the first symptoms appear
	< 6 months	11	18.9	12	20.7	0.788
	≥ 6 months	18	31.1	17	29.3
VEGF-A level
	Low	11	18.9	20	34.5	0.012
	High	18	31.1	9	15.5
LDH level
	Low	8	13.8	18	31.1	0.008
	High	21	36.2	11	18.9

LDH – lactate dehydrogenase, VEGF – vascular endothelial growth factor

To preserve statistical power and adhere to the events- per-variable recommendations for our sample size (n = 58), variables were carefully curated for the final multivariable binary logistic regression model. Gender was excluded from the final multivariable structure to reduce model overfitting and focus strictly on age and downstream bio- logical markers. In this optimised multivariable model, elevated baseline LDH levels (> 66.485 U/l) emerged as a powerful, statistically significant independent predictor of therapeutic failure, conferring a nearly fivefold increase in the odds of a negative response (aOR = 4.684, 95% CI: 1.463–15.000, p = 0.009). On the other hand, VEGF-A and age variable lost the statistical significance in the multivariable analysis (Table 3).

Table 3.

Binary logistic regression for treatment response according to several variables

Parameters	B	SE	aOR	95% CI	p-value
Age (> 45 years)	1.152	0.605	3.164	0.966–10.361	0.057
VEGF-A (> 128.525)	0.722	0.637	2.058	0.590–7.175	0.257
LDH (> 66.485)	1.544	0.594	4.684	1.463–15.000	0.009*
Constant	–1.311	0.528	0.270	–	0.013

aOR – adjusted odd ratio, β – β coefficient, SE – standard error

Discussion

The present study demonstrated that the general characteristics of the patients were similar to those in previous studies. The patient population was predominantly male and above 45 years old. The patients are usually unaware of the symptoms of NPC, even if the neck lumps had already enlarged. Consequently, most patients consulted a physician only after symptoms had persisted for more than six months, resulting in many cases being diagnosed at an advanced stage.

In the present study, elevated serum VEGF-A and LDH levels were significantly associated with a poor response to neoadjuvant chemoradiotherapy in patients with WHO type 3 LA-NPC. Furthermore, multivariate analysis identified LDH as an independent predictor of response (aOR 4.68 with p value of 0.009). These findings suggest that angiogenic activity and tumour metabolic status play a critical role in determining therapeutic response in this population [18–20].

In this study, patients with a negative treatment response had significantly higher median VEGF-A levels compared with responders, and a high VEGF-A level was associated with a more than threefold increased risk of poor response in bivariate analysis. This result is consistent with previous studies reporting that VEGF-A overexpression correlates with an advanced tumour stage, nodal involvement, and inferior treatment outcomes in NPC [11, 21]. Vascular endothelial growth factor A-driven angiogenesis leads to the development of structurally abnormal tumour vasculature, resulting in heterogeneous perfusion and persistent intra-tumoral hypoxia [22]. Such hypoxic conditions are widely known to reduce the effectiveness of radiotherapy, which relies on oxygen to induce DNA double-strand breaks. While tissue hypoxia was not directly quantified in the present cohort, this established radiobiological mechanism provides a highly plausible explanation for the poorer response observed in patients with high baseline serum VEGF-A levels [23].

Although VEGF-A was not retained as an independent predictor in multivariate analysis, this finding does not negate its biological relevance. A vascular endothelial growth factor A protein level is closely linked to other hypoxia- related and metabolic factors, particularly LDH, which may attenuate its independent statistical effect when analysed concurrently [11, 24]. This interaction suggests that VEGF-A may contribute indirectly to treatment resistance through its role in shaping a hypoxic and metabolically adverse tumour microenvironment rather than acting as a solitary determinant of response [22, 25].

Serum LDH levels were significantly higher in patients with negative response, and high LDH remained an independent predictor of poor response after adjustment for confounding variables. This finding aligns with prior studies demonstrating that elevated LDH is associated with an increased tumour burden, higher risk of distant metastasis, and reduced survival in NPC [13, 26, 27]. Lactate dehydrogenase reflects enhanced glycolytic activity characteristic of the Warburg effect, which allows tumour cells to generate energy efficiently under hypoxic conditions and confers resistance to radiotherapy and chemotherapy [28].

According to a pathophysiological perspective, elevated LDH is considered an indirect proxy for increased lactate production and accumulation within the tumour microenvironment. Lactate-induced acidification promotes tumour invasion, suppresses cytotoxic T-cell activity, and stabilises hypoxia-inducible factor-1 α, further enhancing angiogenesis and metabolic adaptation [13, 28, 29]. This theoretical mechanistic pathway provides a coherent, literature- supported framework for why LDH emerged as a stronger and independent predictor of treatment response compared with VEGF-A in the present study.

The observed association between male sex and poor treatment response in bivariate analysis is also noteworthy and consistent with epidemiological data showing worse outcomes among male NPC patients [2, 3]. Hormonal influences, lifestyle-related factors such as smoking, and differences in immune response have been proposed as contributing factors, although these variables were not directly assessed in this study. The loss of statistical significance in multivariate analysis suggests that biological tumour characteristics, particularly metabolic status as reflected by LDH, may outweigh demographic factors in determining treatment response.

Taken together, the results of this study support a model in which hypoxia-driven angiogenesis and metabolic reprogramming jointly contribute to chemoradiotherapy resistance in LA-NPC. High VEGF-A levels indicate active but inefficient angiogenesis, while elevated LDH reflects metabolic flexibility and hypoxia tolerance. The coexistence of these features characterises a tumour phenotype that is less responsive to standard cytotoxic treatment modalities [30].

Clinically, a baseline total serum LDH level serves as an independent prognostic indicator of therapeutic non-responsiveness. Lactate dehydrogenase is a widely available, inexpensive laboratory parameter that can be readily incorporated into routine clinical practice. Measurement of LDH before or during treatment may assist clinicians in identifying patients at high risk of suboptimal response, who may benefit from closer monitoring, treatment intensification, or enrolment in clinical trials evaluating novel therapeutic strategies.

The findings of this study also provide a rationale for exploring combination treatment approaches that target angiogenesis and tumour metabolism. Anti-VEGF therapies have shown potential in normalising tumour vasculature and improving oxygenation, thereby enhancing radiosensitivity, while metabolic inhibitors targeting glycolysis may disrupt the adaptive advantages conferred by high LDH activity [31, 32]. Although the interaction of LDH and VEGF-A was not demonstrated in this study, a previous study hypothesised that both biomarkers may have the synergistic role in NPC response to therapy [33]. Therefore, future prospective studies are needed to validate these findings in NPC.

This study has several limitations that warrant careful consideration when interpreting our findings. First, our case-control design and relatively small sample size (n = 58) from a single-centre setting may restrict generalisability of our data to broader populations and introduces a possibility of selection bias. Second, a major and critical limitation of this study is the presence of residual unadjusted confounding due to the omission of several gold-standard prognostic variables in LA-NPC. Crucially, our dataset lacked records for baseline plasma EBV DNA levels, specific T and N classifications, nodal burden, total tumour volume, baseline performance status (such as ECOG score), smoking history, nutritional metrics (including baseline albumin and haemoglobin levels), and precise metrics of treatment adherence. Because our multivariable binary logistic regression was strictly restricted to three pre-specified variables (age, VEGF-A, and total LDH) to satisfy the events-per-variable criteria and maintain model stability, adjusting for these vital clinical confounders was statistically unfeasible. Consequently, the observed independent predictive value of baseline total serum LDH (aOR = 4.68; p = 0.009) must be interpreted with extreme caution. It represents an independent association strictly relative to the age and VEGF-A parameters evaluated, rather than absolute clinical independence. These thresholds should be viewed as exploratory, and larger, multi-centre prospective cohorts that fully integrate comprehensive clinicopathological and molecular variables are mandatory to validate the true, unconfounded prognostic utility of total serum LDH protein levels. Third, the biomarker measurements were performed at a single time-point prior to treatment, preventing assessment of dynamic, longitudinal trajectories during treatment. Fourth, there remains inherent uncertainty regarding total serum LDH assay interpretation. Total serum LDH lacks isoform specificity; thus, elevations could reflect systemic inflammatory processes or subclinical tissue damage rather than isolated tumour metabolic reprogramming. Fifth, this study relied entirely on immediate short-term therapeutic response metrics and lacked survival outcomes (such as progression-free or overall survival data), which are critical for confirming long-term clinical utility.

To overcome these limitations, future studies should incorporate larger, multi-centre prospective cohorts with robust external validation to verify the true prognostic value of these variables. Longitudinal biomarker assessment across multiple treatment timelines is necessary to track dynamic metabolic and angiogenic adjustments. Furthermore, future investigations must fully integrate comprehensive clinicopathological profiles, particularly baseline plasma EBV DNA copy numbers and exact TNM volumetric variables into multi-variable frameworks to ensure complete adjustment for clinical confounders. Future protocols should also transition to evaluating long-term survival outcomes alongside short-term treatment responses. Finally, investigating more precise biochemical assays, such as specific LDH isoforms (e.g., LDH-5) or combining total serum LDH and VEGF-A into composite, non-invasive molecular risk-stratification panels, represents a highly promising avenue to optimise personalised treatment intensification strategies in LA-NPC.

Conclusions

This study demonstrates that higher pre-treatment baseline serum VEGF-A and total LDH concentrations are significantly associated with a poor therapeutic response 8 weeks post chemoradiotherapy in patients with WHO type 3 LA-NPC. Within our pre-specified, parameter-restricted multivariable model, elevated baseline total LDH (> 66.5 U/l) emerged as a significant predictor of poor treatment response, adjusted for age and VEGF-A. However, because these preliminary findings are derived from a modest, single-centre sample without formal internal validation, calibration assessment, or clinical utility testing, they are strictly exploratory and cannot be used to guide personalised treatment planning or alter standard clinical regimens at this stage.

Disclosures

1. Institutional review board statement: This study was approved by the local ethics committee (approval decision no. 602/UN4.6.4.5.31/PP36/2024, dated: 07.08.2024).

2. Assistance with the article: The University of Mataram provided a grant for the publication.

3. Financial support and sponsorship: None.

4. Conflicts of interest: None.

5. Patient consent: Written informed consent was obtained from the patient for publication.