Baseline ¹⁸F-FDG PET/CT Biomarkers for Prognostication in Pulmonary and Extrapulmonary Small Cell Carcinomas

INTRODUCTION

Small cell lung cancer (SCLC) accounts for 10–15% of lung cancers and is characterised by rapid proliferation, early dissemination, and high recurrence despite initial chemosensitivity.^[1] Extrapulmonary small cell carcinoma (EPSCC) is rare but shares similar histologic and clinical features.^[2,3] Most data on EPSCC come from single-institution experiences and case series, highlighting the need for comprehensive analyses.^[4] Conventional staging systems provide limited prognostic stratification. Novel analytical frameworks, including machine learning–based approaches, have been proposed to improve risk stratification in SCLC cohorts.^[5] F-18 fluorodeoxyglucose (FDG) PET/CT offers functional assessment of tumour biology. While SUVmax reflects focal uptake, volumetric parameters such as metabolic tumour volume (MTV) and total lesion glycolysis (TLG) better represent whole-tumour burden. This study evaluates baseline PET/CT biomarkers for prognostication in SCLC and EPSCC.

MATERIAL AND METHODS

Treatment and response

Patients having Extensive Stage (ES) SCLC or EPSCC underwent combined chemotherapy with 6 cycles of EP regimen (100 mg/m² etoposide, days 1–3, plus 75 mg/m²cisplatin, day 1) every 3 weeks. Limited Stage (LS)-SCLC and selected EPSCC received concurrent chemoradiation; three LS-EPSCC patients underwent surgery plus chemotherapy. LS-SCLC patients with complete response (CR) or partial response (PR) after chemoradiation received prophylactic cranial irradiation (25– 30 Gy in 10–15 fractions). Responses were evaluated using RECIST 1.1 with PET/CT every three cycles, classified as CR, PR, progressive disease (PD), or stable disease (SD). Minimum follow-up was 1 year. The primary endpoint was PFS, defined from baseline PET/CT to progression, recurrence, or death.

Statistical analysis

Descriptive statistics (median, IQR, frequency, range) summarised baseline features [Tables 1 and 2]. Range values are reported in the results text for interpretability, while tables use IQR for statistical consistency. Shapiro–Wilk tested normality. The PFS/OS (whichever was achieved) was the outcome variable of interest. Predictors included demographics, lactate dehydrogenase (LDH), Ki-67, TNM stage, and PET parameters (SUVmax/SUVmean, MTV, TLG at different levels).Mann– Whitney U compared groups. Variables with skewness ≥ 0.9 were log-transformed using the natural logarithm. Cox regression identified independent predictors of PFS [Tables 3 and 4]. ROC curves defined optimal cutoffs and Kaplan–Meier estimated survival. For SCLC patients, representative patient image [Fig 1], Cox regression Forest plot [Fig. 2], ROC curve [Fig. 3] and survival curve [Fig. 4] were generated. Similarly, for EPSCC patients, representative patient image [Fig 5], Cox regression Forest plot [Fig. 6], ROC curve [Fig. 7] and survival curve [Fig. 8] were plotted. Significance was set at p < 0.05. Analyses were performed using SPSS v30 (IBM).

Table 1: Patient characteristics of the SCLC subset

Category	Variable	Value / n (%)
Demographics	Age (years) mean ± 2SD	59.1 ± 15.3
	Gender	Female	6 (15.4%)
		Male	33 (84.6%)
	Smoking	Non-smoker	11 (28.2%)
		Smoker	28 (71.8%)
	Outcome (Progression/ Death)	Not Achieved	2 (5.1%)
		Achieved	37 (94.9%)
TNM Staging	T stage	T1	3 (7.7%)
		T2	2 (5.1%)
		T3	5 (12.8%)
		T4	29 (74.4%)
	N stage	N0	2 (5.1%)
		N1	3 (7.7%)
		N2	12 (30.8%)
		N3	22 (56.4%)
	M stage	M0	5 (12.8%)
		M1	34 (87.2%)
	TNM Stage	IIIA	2 (5.1%)
		Stage IIIB	3 (7.7%)
		Stage IVA	5 (12.8%)
		Stage IVB	29 (74.4%)
	VALSG Stage	LS	7 (17.9%)
		ES	32 (82.1%)
	LDH level	Low	9 (23.1%)
		High	30 (76.9%)
Metastatic sites	LNs	Ipsilateral hilar	14 (35.9%)
		Ipsilateral Mediastinal	17 (43.6%)
		Bilateral hilar/ mediastinal	30 (76.9%)
		Internal mammary	3 (7.7%)
		Supraclavicular	15 (38.5%)
		Other cervical	9 (23.1%)
		Abdominal/ Pelvic	6 (15.4%)
		Retroperitoneal	5 (12.8%)
	Pleural involvement (thickening/ deposits/ effusion)	24 (61.5%)
	Pericardial effusion/ involvement	11 (28.2%)
	Contralateral lung	6 (15.4%)
	Adrenal glands	12 (30.8%)
	Liver	13 (33.3%)
	Bones	13 (33.3%)
	Pancreas	3 (7.7%)
	Kidneys	2 (5.1%)
	Omentum	1 (2.6%)
	Spleen	1 (2.6%)
	Skeletal muscle	1 (2.6%)
	Adnexa	1 (2.6%)
Quantitative biochemical and PET-CT parameters -median (IQR)	LDH (U/L)	288.0 (245.0-532.5)
	Ki67 (%)	80.0 (50.0-90.0)
	Primary tumor size (cm)	7.0 (5.0-10.0)
	PFS (days)	184.0 (117.5-253.5)
	pSUVmax	9.9 (7.8-11.2)
	pSUVmean	3.1 (2.3-4.0)
	pMTV	446.0 (203.7-666.0)
	pTLG	1350.1 (664.4-2381.9)
	tMTV	863.8 (473.5-1287.0)
	tTLG	2067.1 (987.9-4158.1)
	etMTV	305.4 (0.0-885.2)
	etTLG	689.4 (0.0-1973.7)
	wbMTV	1293.0 (899.0-2680.9)
	wbTLG	3772.0 (2077.9-6355.7)

VALSG: Veterans administration lung cancer study group; PFS: Progression-free survival; ES: Extensive stage; LS: Limited stage; LNs: Lymph nodes; p: Primary; t: Thoracic; et: Extra thoracic; wb: Whole body; MTV: Metabolic tumour volume; TLG: Total lesion glycolysis; IQR: Interquartile range, 25th - 75th; LDH: Lactate dehydrogenase; SUVmax: Maximum standardised uptake value; SUVmean: Mean standardised uptake value; IQR: Interquartile range ; T: Tumour; N: Node; M: Metastasis; SD: Standard deviation

Table 2: Primary sites and patient characteristics of EPSCC

Primary Site	Frequency	Percentage (%)	Metastatic Site	Count	Percentage (%)
Urinary bladder	3	14.3%	Regional LNs	17	81.0
Gallbladder	2	9.5%	Liver	10	47.6
Vallecula	2	9.5%	Lung	9	42.9
LN primary	2	9.5%	Bone	7	33.3
Nasal cavity	1	4.8%	Brain	2	9.5
Base of tongue	1	4.8%	Adrenal	2	9.5
Anal canal	1	4.8%	Pleur	1	4.8
Epiglottis	1	4.8%	Peritoneum	1	4.8
Right tonsil*	1	4.8%	None	2	9.5
Cervix	1	4.8%
Prostate	1	4.8%
Esophagus	1	4.8%
Kidney	1	4.8%
GE junction	1	4.8%
Left Pyriform sinus	1	4.8%
Total	21	100%
Patient characteristics
Clinical characteristic	No. (%)
Age (mean ± 2SD)	57.57 ± 22.1 7
Ki67	66.9 ± 23.5
Gender
Male	17 (80.95%)
Female	4 (19.05%)
Smoking
Smoker	9 (42.86%)
Non-smoker	12 (57.14%)
Outcome (Progression/Death)
Achieved	19 (90.48%)
Not achieved	2 (9.52%)
T stage
T1	2 (9.52%)
T2	4 (19.05%)
T3	3 (14.29%)
T4	12 (57.14%)
N stage
N0	4 (19.05%)
N1	1 (4.76%)
N2	5 (23.81%)
N3	11 (52.38%)
M stage
Absent	12 (57.14%)
Present	9 (42.86%)
TNM Stage
Stage II	4 (19.05%)
Stage III	3 (14.29%)
Stage IV	14 (66.67%)
VALSG Stage
LS	9 (42.86%)
ES	12 (57.14%)
LDH
High	13 (61.90%)
Low	8 (38.10%)
Other clinical and PET-CT metabolic parameters -median (IQR)
PFS	220.0 (106.0-277.0)
LDH	276.0 (240.0-425.0)
Ki67	75.0 (40.0-90.0)
pSUVmax	9.7 (7.7-13.3)
pSUVmean	2.8 (2.1-4.1)
pSize	3.4 (2.5-4.4)
pMTV	88.6 (49.6-239.9)
pTLG	244.5 (127.5-1037.9)
prMTV	228.9 (112.9-353.6)
prTLG	507.5 (217.6-1236.9)
eprMTV	0.0 (0.0-1111.0)
eprTLG	0.0 (0.0-1828.9)
wbMTV	360.8 (123.4-1818.5)
wbTLG	1037.9 (217.9-3445.7)

*Presented as carcinoma unknown primary CUP); GE Junction = Esophagogastric junction; LN= Lymph node; PFS: Progression free survival; ES: Extensive stage; LS: Limited stage; p: Primary; pr: Primary region; EPR: Extra primary region; wb: Whole body; MTV: Metabolic tumour volume; TLG: Total lesion glycolysis; IQR: Interquartile range, 25th - 75th; T: Tumour; N: Node; M: Metastasis; VALSG: Veterans administration lung cancer study group

Table 3: Predictors of PFS by univariate and multivariate Cox regression analysis in the EPSCC subset

Parameter	Univariate Analysis				Multivariate Analysis
	B	OR	HR (95% CI: lower–upper)	p value	B	Odds Ratio	HR (95% CI: lower–upper)	p value
Age	0.016	1.02	0.98 – 1.05	0.421
log_LDH	1.668	5.30	1.98 – 14.15	<0.001**	-0.128	0.880	0.225 – 3.445	0.855
Ki67	-0.002	0.998	0.98 – 1.02	0.852
pSize	0.335	1.40	1.04 – 1.87	0.024*
log_pSUVmax	0.704	2.02	0.87 – 4.70	0.102
log_pSUVmean	0.502	1.65	1.05 – 2.61	0.030*	1.067	2.905	0.809 – 10.432	0.102
log_pMTV	0.768	2.15	1.148 – 4.042	0.017*
pTLG	0.001	1.001	1.00 – 1.00	0.015*
log_prMTV	1.523	4.58	1.94 – 10.86	<0.001**
prTLG	0.002	1.002	1.001-1.003	<0.001**
log_wbMTV	1.618	5.05	2.45 – 10.37	<0.001**	2.232	9.314	2.821 – 30.756	<0.001**
log_wbTLG	1.289	3.63	2.024 – 6.511	<0.001**
eprMTV	0.001	1.001	1.00 – 1.002	<0.001**
eprTLG	0.000	1.000	1.000 - 1.000	0.002*
Gender	-0.376	.686	0.223 – 2.112	0.512
Smoking	0.600	1.82	0.72 – 4.62	0.206
T	0.295	1.34	0.87 – 2.08	0.186
N stage 3 vs 1-2	0.531	1.700	.678 – 4.266	0.258
M stage	1.313	3.72	1.39 – 9.93	0.008*
Stage IV vs lower	0.001	1.00	0.51 – 1.94	0.880
VALSG	0.981	2.67	1.01 – 7.02	0.047*	-1.096	0.334	0.064 – 1.757	0.196

VALSG: Veterans administration lung cancer study group; CI: Confidence interval OR - Odds ratio; epr: Extra primary region; LDH: Lactate dehydrogenase; p: Primary; t: Thoracic; et: Extra thoracic; wb: Whole body; MTV: Metabolic tumour volume; TLG: Total lesion glycolysis; * = p value <0.05; **p value < 0.01; T: Tumour; N: Node; M: Metastasis; B: Beta coefficient

Table 4: Predictors of PFS in SCLC cohort by univariate and multivariate Cox regression analysis

Parameter	Univariate analysis				Multivariate analysis
	B	OR	HR (95% CI: lower–upper)	p value	B	OR	HR (95% CI: lower–upper)	p value
Age	-0.036	0.964	0.919 - 1.012	0.141
log_LDH	1.103	3.014	1.624 - 5.593	<0.001**	1.02	2.77	1.30 – 5.91	0.01*
Ki67	0.015	1.015	1.001 - 1.030	0.035*	0.01	1.01	1.00 – 1.03	0.15
pSize	0.144	1.155	1.018 - 1.310	0.026*	0.04	1.05	0.91 – 1.20	0.54
log_pSUVmax	0.700	2.015	0.712 - 5.700	0.187
log_pSUVmean	0.774	2.169	0.974 - 4.830	0.058
log_pMTV	0.988	2.687	1.602 - 4.506	<0.001**
log_pTLG	0.873	2.394	1.544 - 3.714	<0.001**
log_tMTV	1.198	3.313	1.842 - 5.956	<0.001**	0.74	2.10	1.10 – 4.00	0.02*
log_wbMTV	0.568	1.764	1.228 - 2.535	0.002**
log_tTLG	1.134	3.1084	1.842 - 5.246	<0.001**
log_wbTLG	0.643	1.902	1.352 - 2.678	<0.001**
etMTV	0.000	1.000	0.999 - 1.000	0.402
etTLG	0.000	1.000	1.000 - 1.000	0.613
Gender	0.356	1.428	0.588 - 3.464	0.431
Smoking	0.283	1.327	0.640 - 2.751	0.446
T stage IV vs I-III	0.738	2.093	0.952 - 4.599	0.066
N stage	0.661	1.937	0.946 - 3.969	0.071
M stage	0.662	1.940	0.675 - 5.579	0.219
Stage IVB vs lower	0.663	1.940	0.675 - 5.579	0.219
VALSG stage ES vs LS	0.411	1.509	0.623 - 3.655	0.361

VALSG: Veterans administration lung cancer study group; PFS: Progression free survival ES: Extensive stage; LS: Limited stage; N: Lymph node; p: Primary; t: Thoracic; et: Extrathoracic; wb: Whole body; MTV: Metabolic tumour volume; TLG: Total lesion glycolysis; CI: Confidence interval; OR: Odds ratio; HR: Hazard ratio; LDH: Lactate dehydrogenase; * = p value <0.05, **= p value <0.01; B: Beta parameter; T: Tumour; N: Node; M: Metastasis

Close

Fig 1: A 57-year-old man with biopsy-proven small cell lung carcinoma. (A) FDG PET/CT MIP; (B) Coronal (yellow and blue arrows); (C-J) Axial fused PET/CT and axial CECT demonstrate a metabolically active left upper lobe mass with mediastinal and pericardial invasion, with metastatic involvement of mediastinal, left supraclavicular (blue arrows in C and D), cervical level IV (yellow arrows in C and D), perigastric (pink arrows in I and J) lymph nodes, left pleura (white arrows in I and J), and left adrenal gland (green arrows in I and J). The patient progressed after EP chemotherapy and died within 2 month; FDG PET/CT: Fluorodeoxyglucose positron emission tomography/computed tomography; MIP: Maximum intensity projection; CECT: Contrast-enhanced computed tomography; EP: Etoposide+ cisplatin chemotherapy

Export to PPT

Close

Fig 2: Forest plot of multivariate Cox proportional hazards model for progression-free survival (PFS) in SCLC patients. LDH: Lactate dehydrogenase; MTV: Metabolic tumour volume; CI: Confidence interval; t: Thoracic; SCLC: Small cell lung cancer; HR: Hazard ratio

Export to PPT

Close

Fig 3: ROC curve demonstrating superior predictive performance of tMTV (AUC = 0.84) compared to LDH (AUC = 0.74) for detection of shorter PFS (less than the median 184 days) in SCLC patients. tMTV: thoracic MTV, LDH: Lactate dehydrogenase, SCLC: Small cell lung cancer; AUC: Area under curve; PFS: Progression free survival; ROC: Receiver operating characteristic curve

Export to PPT

Close

Fig 4: Kaplan–Meier survival curves for progression-free survival (PFS) of the groups according to the cutoff values derived from the ROC curve (A) 852 cm3 for MTV and (B) 854 IU/mL for LDH in the SCLC cohort. LDH: Lactate dehydrogenase; MTV: Metabolic tumour volume; SCLC: Small cell lung cancer; ROC: Receiver operating characteristic curve; PFS: Progression free survival

Export to PPT

Close

Fig 5: A 65-year-old man presenting with abdominal pain for 1 month. (A) FDG PET/CT MIP (yellow arrow); (B-J) Fused PET/CT coronal and axial and axial CECT demonstrate a metabolically active mass arising from the fundus and body of the gallbladder (yellow arrow in B) with infiltration into adjacent liver parenchyma and metastatic involvement of portocaval (blue arrows in G and H) and peripancreatic (white in I and J) lymph nodes. Biopsy confimed small cell carcinoma. The patient received six cycles of EP chemotherapy, showed an initial response, and progressed after 3 months. FDG PET/CT: Fluorodeoxyglucose positron emission tomography/computed tomography; MIP: Maximum intensity projection; CECT: Contrast-enhanced computed tomography; EP: Etoposide+ cisplatin

Export to PPT

Close

Fig 6: Forest plot displaying hazard ratios (HR) and 95% confidence intervals (CI) from the final Cox proportional hazards model for the EPSCC subgroup. LDH: lactate dehydrogenase; VALSG: Veterans administration lung cancer group; SUVmean: Mean standardised uptake value

Export to PPT

Close

Fig 7: ROC curve analysis of whole-body metabolic tumor volume (wbMTV) for prediction of shorter PFS (less than median 220 days) in EPSCC cohort. EPSCC: Extra pulmonary small cell cancer; ROC: Receiver operating characteristic curve; AUC: Area under curve; wb: Whole body; MTV: Metabolic tumour volume

Export to PPT

Close

Fig 8: Kaplan–Meier curve for PFS in EPSCC patients stratified by wbMTV cutoff of 551.6 cm³, showing significantly longer PFS in patients ≤551.6 compared with >551.6 (log-rank p < 0.001). EPSCC: Extra pulmonary small cell cancer; ROC: Receiver operating characteristic curve; PFS: Progression free survival; AUC: Area under curve; wb; Whole body; MTV: Metabolic tumour volume

Export to PPT

The study was conducted in accordance with the ethical standards of the respownsible institutional committee and with the Declaration of Helsinki (as revised in 2000). This observational research did not qualify as a clinical trial and, therefore, was not registered in the CTRI. Institutional ethical approval was obtained, and written informed consent was obtained from all participants prior to inclusion in the study. Patient confidentiality was strictly maintained.

RESULTS

DISCUSSION

EPSCC is underrepresented in clinical studies, and prognostic PET-based markers have not been systematically evaluated in this rare disease. Despite differing anatomical distributions, SCLC and EPSCC share a glycolytic phenotype with elevated glucose metabolism. This study applied a uniform PET/CT–based analytical framework across both groups, testing the hypothesis that small cell histology, independent of site, expresses a common metabolic behaviour that FDG PET/CT can quantify to provide novel insights into EPSCC tumour metabolism and progression risk, similar to SCLC. The findings support this hypothesis and also reveal differences in the optimal prognostic parameter between pulmonary and extrapulmonary variants. In SCLC, thoracic disease predominated, and log_tMTV emerged as the most powerful prognostic marker. By contrast, EPSCC presented with varied primary sites and widespread dissemination, making log_ wbMTV more representative of the overall disease burden.

Tumour size, a traditional surrogate for disease extent, was significant only in univariate analysis for SCLC and non-predictive in EPSCC, underscoring its limited role in risk stratification. Volumetric PET metrics, by capturing both metabolic activity and viable tumour load, provide a more accurate estimate of biological aggressiveness. This is consistent with prior work showing that MTV and TLG outperform SUVmax and anatomical size in prognostication across several malignancies, including SCLC, NSCLC, and extrapulmonary tumours.^[6–9] SUVmax, while widely reported as an adverse marker, reflects only a single voxel, is subject to partial volume effects, and fails to capture heterogeneity.^[6–8]In contrast, MTV quantifies the entire metabolically active tumour volume, and TLG incorporates both volume and uptake intensity, integrating burden with glycolytic activity.

The current results align with and also extend the prior literature. Park et al. demonstrated prognostic significance of volumetric PET parameters in SCLC, while Hyun et al. did so in NSCLC. Nie et al., in a systematic review and meta-analysis, established MTV and TLG as consistent markers of survival in SCLC. Ong et al. and Zhu et al. validated MTV as an independent predictor of outcome in SCLC in separate studies.^[10–14]

In the SCLC subgroup, log-transformed tMTV emerged as the optimal predictor in Cox regression, with patients above the ROC-derived cutoff of 852 cm³ (based on raw, non–log-transformed tMTV values) showing significantly shorter PFS. Although log_tTLG demonstrated a similar odds ratio, log_tMTV was prioritised in the multivariate model for parsimony and reduced collinearity. Higher log_LDH also predicted worse outcomes, consistent with its established role as a systemic marker of tumour proliferation and necrosis.^[15–17] ROC analysis showed that LDH had high specificity but limited sensitivity, suggesting that while it may serve as a complementary marker, PET-derived volumetric parameters provide stronger risk stratification.

In EPSCC, log_wbMTV demonstrated the strongest prognostic association in Cox regression, with patients above 551.6 cm³ (based on raw, non–log-transformed wbMTV values) experiencing significantly shorter PFS. While log_ wbTLG and prMTV also showed significant univariate associations, multivariate analysis retained log_wbMTV as the most stable parameter, given its broader representation of total disease burden and lower collinearity. This observation is biologically plausible, as EPSCC is a heterogeneous disease arising from multiple sites with varied dissemination patterns, where whole-body measures are needed to capture the full extent of tumour activity. LDH and VALSG stage were significant in univariate analysis but were not retained in multivariate modelling, further emphasising the dominance of log_wbMTV as a prognostic biomarker in this rare entity.

The findings also highlighted shared and distinct biological features of SCLC and EPSCC. In SCLC, the high thoracic disease burden explains the predictive strength of log_tMTV, while in EPSCC, the distributed nature of disease across multiple organ systems makes log_wbMTV a superior predictor. Despite these differences, both cohorts demonstrate the overarching prognostic value of volumetric PET parameters over conventional clinical and pathological markers.

Although many patients with SCLC had FDG-avid extrathoracic disease, whole-body MTV (wbMTV) was not an independent predictor of progression-free survival. This likely reflects the consistently high intrathoracic tumour burden seen in most patients due to late presentation. In SCLC, intrathoracic disease often drives early treatment failure through airway compromise, mediastinal involvement, and locoregional progression, exerting a stronger impact on PFS than total whole-body tumour burden. Accordingly, thoracic volumetric parameters such as tMTV and tTLG may better represent the biologically relevant disease burden in SCLC, even in the presence of extensive extrathoracic metastases.

Ki-67 index is a widely used biomarker in neuroendocrine tumours.^[18–20] In this study, it was not prognostic, highlighting the superior predictive value of PET-derived metrics. This may be explained by uniformly high proliferation indices (>60%) across most cases, which limited its discriminative ability. Similarly, demographic and clinical factors, including age, sex, and smoking history, were not predictive of PFS in either cohort. These observations underscore that conventional clinico-demographic variables fail to capture the biological heterogeneity of small cell carcinomas. In contrast, functional imaging provides a more relevant, site-agnostic assessment of tumour aggressiveness.

Clinically, these results have important implications. In SCLC, high log_tMTV could help identify patients at risk of early progression who may benefit from intensified management, including consolidative thoracic radiotherapy or prophylactic cranial irradiation. In EPSCC, log_wbMTV offers a practical and quantifiable biomarker for prognostication in a disease lacking standardised staging systems, potentially guiding treatment escalation or closer surveillance. Integrating these metrics into clinical decision-making and trial design could enable more precise patient stratification and risk-adapted therapy.

Baseline volumetric PET parameters such as MTV and TLG reflect overall viable tumour burden and biological aggressiveness and may have practical implications for treatment planning in SCLC. Patients with high baseline thoracic MTV or TLG represent a high-risk subgroup prone to early progression. They may benefit from closer surveillance, more frequent response assessment, and consideration of treatment intensification, including consolidative thoracic radiotherapy or timely prophylactic cranial irradiation in appropriate responders. Although prospective validation is required, integrating volumetric PET biomarkers with conventional staging may enable individualised treatment strategies and improved risk stratification in SCLC.

Limitations of this study include its single-centre design and ambispective nature, which may limit generalisability. Some lesions lacked pathological confirmation, though this was mitigated by clinical and imaging follow-up. PET acquisition and reconstruction protocols may vary across centres, potentially influencing quantitative parameters. Nevertheless, the consistent methodology applied here strengthens the internal validity of the results. Future work should aim for multicenter validation with standardised PET protocols and larger patient cohorts. Integrating volumetric imaging biomarkers with molecular or genomic data may further improve prognostic accuracy and pave the way toward personalised treatment strategies.

To conclude, baseline FDG PET/CT–derived volumetric biomarkers demonstrated strong prognostic value in small cell carcinoma. In SCLC, thoracic log_tMTV, along with serum LDH, was an independent predictor of PFS, while in EPSCC, log_wbMTV emerged as the dominant prognostic factor. These findings confirmed that volumetric PET parameters more accurately reflected biological aggressiveness than conventional clinical variables, offering robust tools for risk stratification. Their integration into treatment algorithms and clinical trial design may enable risk-adapted strategies and improve patient management.