Association between insulin resistance biomarkers and metastatic prognosis in treatment-naïve colorectal cancer patients: a pilot study

Abstract

Background

Colorectal cancer (CRC) remains a significant global health burden, ranked among the most common causes of cancer-related fatalities. While insulin resistance (IR) biomarkers have been associated with CRC prognosis, their role in predicting metastasis remains unclear. Metastasis remains a critical determinant of prognosis and treatment planning in CRC. Identifying precise biomarkers can improve CRC management. This study evaluates the prognostic efficacy of lipid-based IR biomarkers in predicting metastasis in treatment-naïve CRC patients and selects the most appropriate one. We also explore their association with clinicopathological characteristics.

Method

Eighty-seven CRC patients (metastatic, n = 24; non-metastatic, n = 63) from four tertiary hospitals in India were analysed. Clinical data included TNM staging, ECOG-PS, KPS, CEA, and lipid profiles. Statistical tests included Fisher’s exact test, Mann-Whitney U-test, ROC curve analysis, Spearman’s correlation, multiple linear regression, and binary logistic regression.

Results

Statistically significant differences were observed in job status, diet, smoking, alcohol use, diabetes, BMI, and IR markers between metastatic and non-metastatic CRC patients. Among the IR biomarkers, the ratio of LDL to HDL (LHR) demonstrated the highest diagnostic accuracy with an AUC of 0.867 (p < 0.05, CI: 0.79–0.94), a sensitivity of 83.3%, and a specificity of 74.6%. Spearman correlation analysis unveiled a moderate-positive relationship between IR biomarkers and carcinoembryonic antigen (CEA) levels, except for the triglyceride glucose index (TyG). Binary logistic regression identified LHR as the sole significant predictor of metastasis, with a one-unit increase in LHR corresponding to a 19.35% higher likelihood of metastasis. Multiple linear regression confirmed a moderate, significant combined effect of TNM staging, ECOG-PS, KPS, and LHR.

Conclusion

LHR strongly predicts metastasis in CRC patients, with high sensitivity and specificity among IR biomarkers. Its significant association with TNM staging, ECOG-PS, and CEA levels highlights its potential for early detection of metastasis and improved risk stratification. Larger studies are needed to validate its clinical utility for personalised treatment planning.

Introduction

According to GLOBOCAN 2022, CRC is the third most common cancer among both sexes [1], with studies indicating that up to 70% of patients develop metastasis or relapse [2–5]. An estimated 178,617 adolescent and young adult cases are expected to occur in India by 2025 [6]. Due to dietary changes, shifting lifestyles and demographic and economic shifts, recent epidemiologic studies have revealed a growing trend of CRC in low and middle-income countries, especially in India [7]. Although several prognostic indicators have been found to predict survival and outcome in patients with CRC, they often fail to predict metastasis. In CRC, metastasis is still crucial in determining prognosis and treatment strategy. Developing a more accurate, cost-effective, and user-friendly biomarker could offer a deeper understanding of the disease course, leading to improved treatment plans and patient outcomes.

Previous studies pointed out that IR has a crucial role in the aetiology of CRC by promoting obesity and metabolic syndrome [8–10]. The metabolic imbalance promoted by IR is the basic mechanism promoting CRC metastasis. Colon, liver and pancreas tumour growth, progression and metastasis are associated with IR, hyperinsulinemia and an increase in IGF-1 [11]. The relationship between IR and lipid-based biomarkers is well established in metabolic dysfunction and cancer pathogenesis, including CRC. Studies have identified the TyG index [12] and HOMA-IR [13] as independent predictors for CRC, and a strong association between IR and CRC in post-menopausal women [14].

Recent studies indicate that tumour cells require major metabolic alterations of substrates, especially lipids, which can boost the metastatic process [15]. These include energetically boosting, triggering signal and epigenetic events, and altering membrane composition to favour metastasis. The advanced stage at presentation, restricted access to healthcare resources and lack of screening result in comparatively higher death rates for CRC in low and middle-income countries [16]. While the non-metastatic CRC 5-year survival rate is 90%, metastatic CRC patient survival is adversely reduced to 12% despite the availability of sophisticated treatment modalities [17]. This highlights a critical shortage of cost-effective prognostic markers, which may hinder the efficacy of targeted cancer therapies. Clinical stratification, prevention, early detection, and medical intervention will substantially decrease the fatality rate of metastatic CRC patients. Identifying prognostic indicators for mCRC is essential for risk reduction; however, while various molecular markers exist, they are costly and require specialised equipment and expertise [18]. In contrast, CEA, which is conventionally used as a prognostic indicator, requires advanced techniques such as chemiluminescence and is not cost-effective. Further, its utility is limited by heterogeneous expression, the influence of nontumor factors and delayed kinetics [19].

The lipid imbalance is documented for several types of cancers, but it still needs to be explored whether lipid parameters are associated with metastasis of CRC. While IR-related biomarkers are linked to CRC pathogenesis and lipid parameters that predict CRC severity, their role in prognosticating mCRC remains largely unclear. Although studies are available on individual IR biomarkers, they do not compare the efficacy of all IR biomarkers in predicting mCRC [20]. Therefore, we aimed to evaluate the prognostic efficacy of lipid-associated biomarkers (LHR, TCHR, TGHR, AI and TyG) in 100 CRC patients and their association with clinicopathological characteristics in predicting mCRC.

Materials and methods

The study was conducted from December 2023 to December 2024 at the All India Institute of Medical Sciences (AIIMS), Rishikesh and data from four participating tertiary cancer centres were compiled. Data collected at the time of initial diagnosis, with no follow-up planned. This study employed a time-bound convenience sampling, and a formal sample size calculation was not performed. The institutional ethics committee approved the study (Reference No. AIIMS/IEC/24/241). This study adheres to the ethical principles outlined in the Declaration of Helsinki [21]. Individuals with metachronous colorectal cancer and low-performance scores were excluded. The inclusion criteria comprised de novo, treatment-naive patients with histopathologically confirmed colorectal cancer (stage I-IV). One hundred treatment-naïve CRC patients were screened, and Fig. 1 shows the procedure for patient selection.

Fig. 1.

Procedures used for the selection of patients with CRC

A total of 13 patients were excluded from the study: five patients had missing fasting lipid profile data, two patients had two primary tumours, and six patients lacked available records before treatment. Consequently, data from 87 patients were analysed. Prospective data compilation was done from patient records as per the case record form after obtaining informed consent. Data included age, height, weight, KPS score, ECOG-PS, alcohol consumption, smoking status, and blood parameters like TC, LDL, HDL, TG and FBS. For TNM staging, the 8th edition of the American Joint Committee (AJCC) cancer staging system was used [22]. The CEA level was measured using the chemiluminescence method on the Siemens ADVIA Centaur Autoanalyser, while the lipid profile was assessed using the Beckman Coulter AU 680 clinical chemistry analyser. Insulin resistance biomarkers were calculated. LHR is defined as the ratio of LDL to HDL, TCHR is the ratio of TC to HDL, and TGHR is the ratio of TG to HDL. Triglyceride glucose index (TyG) = Ln [(TG×FBG)/2], and the atherogenic index is calculated by (TC-HDL)/HDL [23–25]. Additionally, body mass index (BMI) is the ratio of weight (kg) to height (m) squared [26].

Statistical analysis

All statistical analyses were performed using SPSS v27.0 and GraphPad Prism 8. Descriptive statistics were used to summarise categorical variables as proportions and continuous variables as mean ± SD. Fisher’s exact test was applied for categorical variable comparisons, while the Mann-Whitney U-test was used for non-normally distributed continuous data. The diagnostic accuracy of IR biomarkers for detecting metastasis was evaluated using ROC curve analysis, and the area under the curve (AUC), sensitivity, and specificity were reported. Spearman correlation assessed the association between CEA levels and IR biomarkers. Binary logistic regression determined the predictive capability of LHR for metastasis, estimating odds ratios with 95% confidence intervals. Multiple linear regression assessed the combined effect of TNM staging, ECOG-PS, KPS, and LHR on metastatic progression. A p-value < 0.05 was considered statistically significant.

Results

Sociodemographic characteristics of the patients

The major population characteristics of the subjects include age, gender, employment status, dietary habits, smoking status, alcohol consumption, diabetes mellitus, family history and BMI. The mean age of the total patient population was 47.8. Males constituted 63.2%, while females accounted for 36.8% of the cohort. The study cohort was categorised into metastatic CRC (n = 24) and non-metastatic CRC (n = 63); in the metastatic category, liver metastasis was predominant. A total of 13 patients had a history of smoking, while five reported alcohol consumption. Diabetes mellitus was present in eight patients. Labourers and farmers were classified as non-sedentary, while professionals and others were classified as sedentary for job status classification. At the same time, housewives, depending on their daily activities, were included in sedentary or non-sedentary groups. In both groups, most patients followed a mixed diet (vegetarian and non-vegetarian). Table 1 shows detailed sociodemographic characteristics.

Table 1.

Socio-demographic profile of the patients with colorectal cancer

Characteristics		Metastatic colorectal cancer (n = 24), No. (%)	Non-metastatic colorectal cancer (n = 63), No. (%)	p Value*
Age	> 50	9 (37.5)	37 (58.7)	0.095
	< 50	15 (62.5)	26 (41.2)
Gender	Male	14 (58.3)	41 (65.0)	0.622
	Female	10 (41.6)	22 (34.9)
Job-status	Sedentary	20 (83.3)	35 (55.5)	0.024*
	Non-sedentary	4 (16.6)	28 (44.4)
Diet	Non-veg./mixed	20 (83.3)	29 (46.0)	0.001*
	Pure veg.	4 (16.6)	34 (53.9)
Smoking	Smoker	8 (33.3)	5 (7.9)	0.005*
	Non-smoker	16 (66.6)	58 (92.0)
Alcohol	Yes	5 (20.8)	0	0.001*
	No	19 (79.1)	63 (100)
Diabetes	Yes	6 (25)	2 (3.1)	0.004*
	No	18 (75)	61 (96.8)
Family history	Present	0	8 (12.6)	0.100
	Absent	24 (100)	55 (87.3)
BMI	≥ 24	16 (66.6)	20 (31.7)	0.006*
	< 24	8 (33.3)	43 (68.2)

^*Fisher’s exact T-test, p < 0.05, is statistically significant

A statistically significant difference was observed between the groups regarding job status, dietary habits, smoking status, alcohol consumption, diabetes mellitus and BMI. However, no significant differences were found in age, gender or family history. The mean values of LHR, TCHR, TGHR, AI and TyG were significantly higher in the metastatic group. Additionally, all IR biomarkers showed a statistically significant difference between the mCRC and CRC groups (p < 0.05) except AI (Table 2).

Table 2.

Different insulin resistance biomarkers in patients with colorectal cancer

Insulin resistance biomarkers	Metastatic colorectal cancer Mean ± SD (n = 24)	Non-metastatic colorectal cancer Mean ± SD (n = 63)	p Value*	FDR adjusted p-value
LHR	3.305 ± 1.129	1.892 ± 0.732	< 0.000*	0.0005
TCHR	4.548 ± 0.893	3.272 ± 1.023	< 0.000*	0.00025
TyG	3.548 ± 0.893	2.272 ± 1.023	< 0.000*	0.000167
TGHR	3.960 ± 2.265	2.680 ± 1.351	0.001*	0.00125
AI	8.716 ± 0.486	8.649 ± 0.429	0.399	0.399

^*Mann-Whitney test, p < 0.05 is statistically significant

All IR biomarkers except AI remain significant (FDR < 0.05) after Benjamini-Hochberg adjustment

Among the IR biomarkers, LHR demonstrated the highest diagnostic accuracy for metastasis with an AUC of 0.867 (95% CI: 0.79–0.94, p < 0.05), a sensitivity of 83.3%, a specificity of 74.6%, and a cut-off value of 2.24. The diagnostic performance of other biomarkers was as follows: TCHR cut-off value 3.4177 sensitivity 100% specificity 65.1% AUC 0.849, AI cut-off value 2.4177 sensitivity 100% specificity 65.1% AUC 0.849, TGHR cut-off value 2.4241 sensitivity 87.5% specificity 58.7% AUC 0.714 and TyG cut-off value 8.5439 sensitivity 75.0% specificity 49.2% AUC 0.559. There was a similarity between the ROC curves of TCHR and AI; both have an AUC of 0.849. Further details are provided in Fig. 2; Table 3.

Fig. 2.

Comparison of different Insulin Resistance biomarkers’ ROC curves (LHR, TGHR, TCHR, AI and TyG) for detecting metastasis, the ROC curves of TCHR and AI; both have an AUC of 0.849

Table 3.

Area under the curve and 95% confidence interval of IR biomarkers ROC curves

Insulin resistance biomarkers	Area under the curve	95%CI	P Value	FDR adjusted p-value
LHR	0.867	0.790 to 0.944	0.000	0.0005
TGHR	0.714	0.600 to 0.828	0.002	0.0025
TCHR	0.849	0.770 to 0.927	0.000	0.00025
AI	0.849	0.770 to 0.927	0.000	0.000167
TyG	0.559	0.425 to 0.694	0.395	0.395

LHR, TGHR, TCHR and AI remain statistically significant (FDR-adjusted p < 0.05)

Spearman’s rank correlation was performed to assess the strength and direction of the association between IR biomarkers and CEA. A significant moderate positive correlation was observed between LHR, TCHR, AI and CEA. LHR and CEA: r (85) = 0.307, p = 0.004, TCHR and CEA: r (85) = 0.318, p = 0.003, AI and CEA: r (85) = 0.318, p = 0.003. In contrast, no significant correlation was found between TGHR, TyG and CEA. TGHR and CEA: r (85) = 0.162, p = 0.134, TyG and CEA: r (85) = 0.068, p = 0.528 (Fig. 3). The average value of CEA in the mCRC patients was 83.05 ng/mL, and for the non-metastatic group was 16.01 ng/mL (Reference range 0.5 ng/mL to 100 ng/mL).

Fig. 3.

Scatter plots visualising the Spearman’s rank correlation between insulin resistance biomarkers and CEA levels

Binary logistic regression analysis evaluated the association of LHR, TCHR, TGHR, AI and TyG with metastasis. Among these biomarkers, LHR emerged as the only significant predictor of metastasis. When all predictor values (X_j) were zero, the odds of non-metastatic disease compared to metastatic disease were 154.0006, indicating a strong baseline effect. Furthermore, a one-unit increase in LHR was associated with a 19.35% higher likelihood of metastasis, reinforcing its potential as a key predictive biomarker (Table 4; Fig. 4).

Table 4.

Binary logistic regression analysis of insulin resistance biomarkers with metastasis

Predictor	Coefficient (β)	SE	z-stat	95% CI for OR	OR (exp(β))	p-value
Intercept (bo)	5.0370	1.0176	4.9500	(3.0426, 7.0314)	154.0006	<0.001
LHR	–1.6423	0.3830	–4.2874	(–2.3930, − 0.8915)	0.1935	<0.001
TGHR	–0.7764	0.4837	–1.6051	(–1.7244, 0.1717)	0.4601	0.1085
TCHR	29.8688	—	—	—	—	—
AI	–28.5197	—	—	—	—	—
TyG	1.2510	1.2803	0.9771	(–1.2585, 3.7604)	3.4937	0.3285

Fig. 4.

Forest plot for Binary Logistic regression analysis for the association of LHR, TGHR and TyG with metastasis

The association between clinicopathological characteristics (TNM staging, ECOG-PS, KPS) and LHR was evaluated by using multiple linear regression analysis. The results indicated a moderate collective, significant association between clinicopathological characteristics and LHR in predicting metastatic progression. The current study presented LHR as a reliable prognostic predictor for mCRC (Fig. 5).

Fig. 5.

Forest plot for multiple linear regression analysis showing the association between clinicopathological characteristics (TNM staging, ECOG-PS, KPS) and LHR

Discussion

Metastatic CRC is associated with poor survival outcomes. While molecular biomarkers are under consideration, their clinical utility is limited by the need for specialised equipment and high costs. Blood-based markers like CEA are widely used but lack specificity and cost-effectiveness. Lipid-based markers may serve as valuable adjuncts to CEA and imaging techniques. Several studies have linked metabolic reprogramming to CRC risk and progression; however, their role in metastasis remains unexplored. This study investigated the predictive ability of IR biomarkers for detecting metastasis in CRC patients. Our findings revealed that 27.5% had metastatic disease, while 52.8% were younger than 50 years, with a mean age of 47.8 years. Similarly, an audit of 978 CRC cases from the Tata Memorial Hospital [27] reported that 23.5% were metastatic and 26.8% were less than 40 years, with a mean age of 47.2.

A significant difference was observed in job status, diet, smoking status, alcohol consumption, diabetes mellitus and BMI between the metastatic and non-metastatic groups. These sociodemographic characteristics are major risk factors for CRC, due to their role in promoting chronic inflammation, oxidative stress, insulin resistance and mutations. In our study, 20 out of 24 (83.3%) mCRC patients had sedentary lifestyles, 20 (83%) followed non-vegetarian or mixed diets, 8 (33%) were smokers, 5 (20%) consumed alcohol, 6 (25%) had diabetes mellitus and 16 (66.7%) had BMI ≥ 24, all of which were associated with poor prognosis. Considering the total number of patients in our study, the average BMI was 21.15 in non-metastatic CRC patients and 23.29 in mCRC patients, with 36 out of 87 (47.4%) patients having a BMI ≥ 24; additionally, 55 (63.2%) had sedentary jobs, and 49 (56.3%) followed a non-vegetarian or mixed diet. Lower BMI in our patients is associated with higher cachexia percentage and TNM stage III and IV. Non-obese patients in our study with high insulin resistance are likely to have relatively high body fat and are metabolically obese. The possible explanation may be that excess adipose tissue is associated with increased levels of IL-6, TNF α and CRP, which contribute to a tumour-promoting microenvironment and enhance metastatic potential [28]. Additionally, excess adipose tissue dysfunction leads to an imbalance in adipokine secretion, characterised by increased leptin and reduced adiponectin levels, further driving CRC metastasis [29]. Obesity-associated immune suppression leads to reduced cytotoxic T-cell activity and increased myeloid-derived suppressor cells, which enables tumour evasion and metastasis [30].

In our study, high LHR or severe insulin resistance is associated with an increased risk of poor prognosis and metastasis in CRC patients. Notably, the high LHR AUC (0.867) highlights the strong predictive ability of our model for CRC metastasis with a cut-off value of 2.24 and a one-unit increase in LHR corresponding to a 19.35% higher likelihood of metastasis. Therefore, we propose LHR as a simple yet effective lipid-based biomarker for metastasis risk assessment in CRC. A potential explanation for observing higher LHR in our patient population is that increased LDL promotes oxidative stress and inflammation in the colon. Moreover, oxidised LDL induces the production of pro-inflammatory cytokines, which contribute to cancer progression and metastasis [31]. At the same time, insulin resistance exacerbates inflammation by stimulating free radical production [32]. Indians with hyperinsulinemia have low levels of HDL, and high levels of small dense LDL and LP (a) [33]. High LHR is linked to high insulin resistance, which elevates insulin and insulin-like growth factor-1 [34]. These promote cell proliferation, inhibit apoptosis and enhance CRC growth and metastasis. Oxidised LDL generates reactive oxygen species, causing lipid peroxidation, DNA damage, mutation and increased risk of tumour growth and metastasis [35]. Deranged lipid levels, especially LHR, influence gut microbiome composition, favouring bacteria that produce carcinogenic metabolites, such as secondary bile acids that promote CRC tumorigenesis and metastasis [36]. An increased LHR reflects a proinflammatory, prooxidative and metabolically deranged state that promotes CRC and metastasis [37]. Differing findings were also reported as reduced levels of TG, LDL and TyG index in CRC patients with lymph node metastasis, and a combination of TyG index, CEA and CA 19 − 9 yielded the highest positive prediction rate for CRC [38].

This study is the first to examine cholesterol-related IR biomarkers and their role in CRC de novo treatment-naïve patient progression and metastasis prediction. While previous studies have examined individual IR biomarkers, we systematically compared LHR, TCHR, TGHR, TyG, and AI to identify the most effective prognostic marker. The identification of LHR as the most relevant factor is significant, as no prior studies have comprehensively compared these biomarkers for metastasis prediction. This study established a significant association between LHR and prognostic indicators like CEA, ECOG-PS, KPS and TNM staging, confirming its clinical utility. This suggests that LHR serves as a tumour-related biomarker for disease severity, progression and prognosis in mCRC. LHR is derived from standard lipid profiles; it could be rapidly available, cost-effective and beneficial in resource-limited settings where advanced molecular or imaging tools are less accessible. Our findings are consistent with previous research, such as Notarnicola et al. [39], Lieu et al. [40, 41] Rodrigues et al., and Wang H et al. [42], which are individual IR biomarker studies in CRC pathogenesis and progression. Many studies face limitations, as patient medications can confound results. However, our study is novel in that all subjects were treatment-naïve, ensuring that prior medication use did not affect our findings.

Patients with higher LHR values demonstrate poorer clinical performance (ECOG-PS, KPS) and more advanced disease (TNM staging). Identifying high-risk patients early can guide aggressive treatment strategies and enhance patient outcomes. Exploring lipid-lowering therapies, such as statins, PCSK9 inhibitors, or lifestyle modifications, may be beneficial as adjuncts to standard mCRC treatments. Personalised treatment plans could incorporate lipid profiles through precision medicine approaches. The correlation of LHR with CEA indicates that there may be a connection between dyslipidaemia and cancer progression, leading to novel insights into cancer metabolism and lipid-related tumorigenesis. Unlike imaging techniques, LHR measurements are non-invasive, reproducible, and can be tracked longitudinally to monitor disease progression or treatment response.

Despite its strengths, this study has certain limitations. The relatively small sample size, lack of follow-up may limit the generalizability of our findings. Additionally, the cross-sectional design prevents us from establishing a causal relationship between IR biomarkers and metastasis progression. Lifestyle changes, diet and comorbidities might have influenced the results. Although our focus on treatment-naïve patients minimises confounding from prior therapies, it may not fully capture the heterogeneity of CRC across different treatment stages. We hypothesised that temporal variations would have minimal impact on the results. Combinations of IR biomarkers were not analysed in this study due to their exploratory nature and limited sample size, and suggest future studies explore combined biomarker panels to assess potential improvements in diagnostic accuracy. The prognostic predictive efficiency may be enhanced by increasing sample size and a broader range of clinical features and data.

Conclusion

Study highlights the prognostic significance of lipid-based insulin resistance biomarkers in CRC, identifying LHR as a promising and effective predictor of metastasis. LHR is a better prognostic biomarker for mCRC, especially in the Indian setup. Our findings highlight the necessity of reinforcing population awareness about healthy lifestyle habits and point to the need for multicentric longitudinal survival studies with long-term follow-up to validate the prognostic utility of LHR across diverse ethnic populations. Combining LHR with other emerging biomarkers and advanced imaging techniques could refine risk stratification and facilitate more personalised treatment strategies. Furthermore, mechanistic studies examining the relationship between lipid metabolism, insulin resistance and metastatic signalling pathways will shed more light on CRC pathogenesis and progression. Exploring novel therapeutic strategies to enhance insulin sensitivity may reveal innovative approaches to reduce the risk of metastasis in CRC patients.

Abbreviations

CRC

Colorectal cancer

mCRC

Metastatic colorectal cancer

IR

Insulin resistance

ECOG-PS

Eastern Cooperative Oncology Group performance score

KPS

Karnofsky Performance Status

CEA

Carcinoembryonic antigen

LHR

Low-density lipoprotein to high-density lipoprotein ratio

TGHR

Triglyceride to high-density lipoprotein ratio

TCHR

Total cholesterol to high-density lipoprotein ratio

AI

Atherogenic index

TyG

Triglyceride index

IL-6

Interleukin 6

TNFα

Tumour necrosis factorα

CRP

C reactive protein

BMI

Body mass index

FBG

Fasting blood glucose

TC

Total cholesterol

TG

Triglyceride

LDL

Low-density lipoprotein

HDL

High-density lipoprotein

Author contributions

Concept and design: MPN, AS and BG Acquisition, analysis, or interpretation of data: MPN, AS, BG, PK, SB, SG, AG, DS, SWGDrafting of the manuscript: MPN, AS and BG Critical review of the manuscript for important intellectual content: MPN, AS, BG, PK, SB, SG, AG, DS, SWG and AAMSupervision: AS, BG and AAM.

Funding

No funding was received for this study.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Prospective data were compiled from all patient records as per the case record form after obtaining informed consent. The institutional ethics committee approved the study (Reference No. AIIMS/IEC/24/241).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.