Introduction
Colorectal cancer (CRC) is the fourth and third most common cancer in Iran and the world, respectively [
1‐
3]. The incidence and mortality rates of CRC have been increasing in recent years [
4], so that in 2020, CRC accounted for 1.14 million new cases worldwide [
5]. Various factors (modifiable and non-modifiable risk factors) affect CRC development [
6]. Epidemiological studies have shown a relationship between diets and CRC risk [
7‐
9]. Some dietary factors, including carbohydrates, lead to the proliferation of cancer cells, including CRC, through alternations in insulin levels and circulating glucose, impaired glucose metabolism, insulin resistance, and hyperinsulinemia [
10]. The ability of carbohydrates to affect blood glucose and insulin concentrations differs substantially and depends on the diet’s amount, composition, and quality [
11,
12]. Glycemic load (GL), glycemic index (GI), carbohydrate quality index (CQI), and low-carbohydrate diet score (LCDS) are used to assess the quantity and quality of carbohydrates in foods [
13,
14]. The GI indicates how food’s carbohydrate content influences blood glucose levels [
15]. Also, GL offers the impact of all dietary carbohydrates on glucose after a meal [
11]. However, a single component cannot be a suitable criterion for assessing the quality of carbohydrates, so the CQI was introduced as an indicator of dietary carbohydrate quality that consists of the GI and dietary fiber intakes, whole grains, and solid or liquid carbohydrates [
16]. Overall, all data collectively support a central role of glucose metabolism in carcinogenesis and lead to significant interest in LCDS as a practical dietary approach for cancer prevention [
17].
It has been shown that when assessing the insulin response, carbohydrates are not the only stimulus for its release, so the insulin load (IL) and dietary insulin index (II) [
18] were presented [
19]. II indicates the insulin response after a meal, including protein, fat, and carbohydrates, compared to an isoenergetic portion of a reference meal (white bread or glucose). IL is also computed by multiplying each food’s II by its consumption frequency and energy content [
20].
Previous studies have investigated the relationship between GL, GI, IL, and II with CRC risk, and the potential relationship between glucose metabolism and cancer is still debated [
21‐
24]. Furthermore, despite the important role of CQI in assessing the quality of carbohydrates, studies have not yet evaluated the association between CQI and LCDS with CRC odds. To our knowledge, previous research has not yet simultaneously demonstrated the association of overall quality and quantity of carbohydrate intake on the odds of CRC. In conclusion, the findings of this research contribute to understanding the potential relationship between total carbohydrate intakes and CRC odds. Therefore, the current study investigated the association between CQI, LCDS, and other indices (GL, GI, IL, and II) and the odds of CRC.
Results
The baseline characteristics of the study population are presented in Table
1. The mean and SD of the age of the participants in the case and control group were 58.2 ± 10.4 and 57.7 ± 10.4, respectively (P = 0.746). Also, the BMI of the case group was 27.6 ± 4.2 and the control group one was 26.6 ± 4.2 (P = 0.362). There were significant differences in the history of CRC (P = 0.017), taking aspirin (P = 0.016), acetaminophen (P = 0.004), and vitamin/mineral supplementation (P = 0.015) between the two groups. Regarding dietary intake, the intake of protein (P = 0.048) and fiber (P < 0.001) was lower in the case group than in the control group. In contrast, the fat intake (P < 0.001) was significantly higher in the case group than the control group. GI was greater in the case group than in the control group (P = 0.001). However, the total LCDS was higher in the control group than in the case group (P < 0.001). There were no significant differences in II and CQI between the case and control groups (P > 0.05).
Table 1
The basic characteristic of the control (n = 71) and case groups (n = 142)
Age (year) 1 | 58.2 ± 10.4 | 57.7 ± 10.4 | 0.746 |
Energy (kcal/day) 1 | 2262.3 ± 450.1 | 2255.2 ± 341.2 | 0.908 |
Carbohydrate (g/day) 1 | 347.5 ± 89.6 | 354.8 ± 71.8 | 0.552 |
Protein (g/day) 1 | 79.1 ± 17.2 | 83.8 ± 14.3 |
0.048
|
Fat (g/day) 1 | 65.8 ± 8.1 | 60.5 ± 8.4 |
<0.001
|
Fiber (g/day) 1 | 18.9 ± 2.3 | 20.4 ± 3.1 |
<0.001
|
Glycemic index 2 | 63.6 (5.9) | 61.7 (6.6) |
0.001
|
Insulin index 2 | 44.0 (17.4) | 41.7 (14.7) | 0.087 |
CQI total score 2 | 11.0 (4.0) | 13.0 (3.0) | 0.177 |
Total LCDS2 | 34.0 (22.0) | 36.0 (13.2) |
<0.001
|
BMI (kg/m2) 1 | 27.6 ± 4.2 | 26.6 ± 4.2 | 0.362 |
Income (dollar) 2 | 393.0 (253.0) | 402.0 (302.0) | 0.206 |
Physical activity (MET-h/day) 1 | 36.8 ± 3.6 | 36.7 ± 4.8 | 0.932 |
History of CRC, % 3 | | |
0.017
|
Yes | 7 (9.9) | 3 (2.1) | |
No | 64 (90.1) | 139 (97.9) | |
Smoking, % 3 | | | 0.164 |
Never | 57 (80.2) | 101 (70.1) | |
Former | 8 (11.3) | 15 (10.6) | |
Current | 6 (8.5) | 26 (18.3) | |
Ibuprofen, % 3 | | | |
Yes | 5 (7.0) | 22 (15.5) | |
No | 66 (93.0) | 120 (84.5) | |
No | 70 (98.6) | 128 (90.1) | |
Acetaminophen, % 3 | | |
0.004
|
Yes | 4 (5.6) | 28 (19.7) | |
No | 67 (94.4) | 114 (80.3) | |
Taking vitamin and mineral supplements, % 3 | | |
0.015
|
Yes | 8 (11.3) | 35 (24.6) | |
No | 73 (88.7) | 107 (75.4) | |
Table
2 shows ORs and 95% confidence intervals (CIs) in the multivariable-adjusted and crude models across tertile of GI, GL, II, IL, CQI, and LCDS. As can be seen, in the adjusted model, the odds of CRC in the third tertile were significantly higher than the first tertile of GI (OR = 3.89; 95% CI: 1.71–8.84). Also, we found a significant association between the second tertile of GL and the odds of CRC in the adjusted model (OR = 2.42; 95% CI: 1.07–5.47). Moreover, there was a significant association between IL and CRC odds in the second and last teritles compared to the first tertile in the adjusted model (tertile (T)
2-OR = 2.46; CI: 1.08–5.61 and T
3-OR = 2.80; 95% CI: 1.07–7.31). In contrast, there was a negative relationship between CQI and CRC in the adjusted model (T
2-OR = 0.24; 95% CI: 0.11–0.53 and T
3-OR = 0.15; 95% CI: 0.06–0.39). Also, individuals in the second and third tertiles of LCDS had lower odds of CRC than the first tertile (T
2- OR = 0.33; 95% CI: 0.13–0.79 and T
3- OR = 0.28; 95% CI: 0.10–0.82).
Table 2
Crude and multivariable-adjusted OR and 95% CIs across tertile of GI, GL, II, IL, CQI, and LCDS (in 71 cases and 142 controls)
GI
| 71/142 | | |
T1 | 17/54 | 1.00 (Reference) | 1.00 (Reference) |
T2 | 21/50 | 1.33 (0.63–2.81) | 1.35 (0.59–3.04) |
T3 | 33/38 |
2.75 (1.34–5.65)
|
3.89 (1.71–8.84)
|
Ptrend | |
0.005
|
0.001
|
GL
| 71/142 | | |
T1 | 19/52 | 1.00 (Reference) | 1.00 (Reference) |
T2 | 28/43 | 1.78 (0.87–3.62) |
2.42 (1.07–5.47)
|
T3 | 24/47 | 1.39 (0.68–2.87) | 1.85 (0.73–4.70) |
Ptrend | | 0.374 | 0.165 |
II
| 71/142 | | |
T1 | 21/50 | 1.00 (Reference) | 1.00 (Reference) |
T2 | 23/48 | 1.14 (0.56–2.32) | 1.08 (0.48–2.42) |
T3 | 27/44 | 1.46 (0.72–2.94) | 1.82 (0.80–4.13) |
Ptrend | | 0.286 | 0.122 |
IL
| 71/142 | | |
T1 | 18/53 | 1.00 (Reference) | 1.00 (Reference) |
T2 | 27/44 | 1.91 (0.92–3.95) |
2.46 (1.08–5.61)
|
T3 | 26/45 | 1.87 (0.90–3.86) |
2.80 (1.07–7.31)
|
Ptrend | | 0.098 |
0.027
|
CQI
| 71/142 | | |
T1 | 30/32 | 1.00 (Reference) | 1.00 (Reference) |
T2 | 26/65 |
0.34 (0.17–0.68)
|
0.24 (0.11–0.53)
|
T3 | 15/45 |
0.27 (0.12–0.59)
|
0.15 (0.06–0.39)
|
Ptrend | |
0.001
|
<0.001
|
LCDS
| 71/142 | | |
T1 | 30/40 | 1.00 (Reference) | 1.00 (Reference) |
T2 | 18/53 |
0.45 (0.22–0.92)
|
0.33 (0.13–0.79)
|
T3 | 23/49 | 0.62 (0.31–1.24) |
0.28 (0.10–0.82)
|
Ptrend | | 0.173 |
0.021
|
Discussion
Our research indicated a positive association between GI, GL, IL, and CRC odds. Furthermore, we found a negative association between CQI, LCDS, and CRC odds.
A compound of genetic and environmental factors, especially diet, plays a role in cancer etiology [
38,
39]. The current study demonstrated a significant positive association between dietary GL and GI and CRC odds. GI and GL indicate different dimensions of consumed carbohydrates. GI provides information about the overall quality of carbohydrates in the diet. In contrast, dietary GL, which reflects the amount of carbohydrate intake, contains both the quantity and quality of carbohydrate intake in the diet [
40].
A study by Choi et al. revealed that dietary GI has a positive and significant relationship with the risk of CRC [
41]. Moreover, the findings from a prospective study showed that increasing consumption of a high GI diet was significantly related to an increased risk of CRC [
42]. Consumption of a high glycemic diet is equal to high blood glucose levels. The amount of serum insulin increases after glucose rises. This hormone arouses cancer growth by decreasing insulin-like growth factor (IGF) binding protein, increasing the bioactivity of IGF-1, and changing the metabolism of the sex hormone [
11]. Research has shown that high IGF-1 and C-peptide (indicating increased insulin rates) are related to a remarkable increase in CRC risk [
43,
44].
Our findings suggested that higher IL might be associated with higher odds of CRC. IL, which has taken significant consideration in recent years, is a suitable indicator to predict the risk of chronic diseases [
45,
46]. Similar to our observations, a study revealed that higher IL was associated with CRC risk [
47]. In another study, it was observed that higher scores of dietary insulin were associated with a statistically significant increment in mortality after CRC diagnosis [
48]. Furthermore, it has been reported that higher II is significantly related to a higher risk of recurrence and mortality in patients with colon cancer [
49]. In contrast to our results, a prospective study showed that dietary IL were not associated with the risk of CRC [
24].
Also, a significant negative relationship was observed between LCDS and the odds of CRC. Several studies have shown an association between LCD and some cancers [
50‐
52]. A prospective study by Cai et al. indicated that animal-based LCD was related to a greater risk of CRC [
52]. Moreover, Song et al. illustrated that vegetable-based LCD was related to lower CRC-specific mortality [
53]. It has been shown that both hyperinsulinemia and hyperglycemia are associated with a poor prognosis of CRC [
54‐
56]. These studies demonstrated the critical role of glucose metabolism in carcinogenesis and led to considerable interest in LCD as a beneficial dietary approach to help in cancer treatment [
17]. In addition, as mentioned earlier, high carbohydrate intake and high GL may increase blood glucose, and as a result, insulin increases [
18,
57]. Insulin has been shown to stimulate cancer cells, reduce apoptosis, and can increase carcinogenesis through IGF-1 [
58]. As a result, it seems that LCD can reduce the odds of cancer by the mentioned mechanisms.
Also, the findings demonstrated an inverse association between CQI and CRC odds. Despite the important role of CQI in evaluating the quality of carbohydrates, no studies have evaluated the association between CQI and CRC odds. However, an inverse relationship has been observed between higher CQI diets and breast cancer risk [
59], obesity [
60], cardiovascular disease incidence [
33], and all-cause mortality [
61], compared to lower CQI diets. In line with our findings in the study of Sasanfar et al. it was also found that greater CQI scores were related to a lower breast cancer risk [
59].
The strengths of the present study are that despite the significant role of CQI in assessing the quality of carbohydrates, studies have not yet evaluated the association between CQI and LCDS and the odds of CRC with this method. Also, we used valid and reliable questionnaires for data collection [
26], which can further support the accuracy of the findings. In addition, our study has several limitations that should be noted. First, the study sample size was small. Second, the data are case-control, which prevents us from concluding the causal relationships between the variables. Third, we used similar foods for the limited number of foods that II was unavailable. Therefore, further II tests are required to support our results in the present study.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.