Nine studies (24 experiments) that reported the effect of CLA feeding on growth performance and/or FAs profiles in meat of broiler chickens were identified by a computerized literature search. Research articles were selected based on use of CLA, presence of a control group without CLA supplementation and presence of a measurement of standard deviation or error.

The mean difference between control and treatment groups using Hedges’s adjusted g for pooling was used for calculation of a standardized mean difference (SMD). Estimated effect sizes from used studies were visualized in forest plot. Summary effects across the studies were calculated using fixed and random effects models. The inverse weighting for pooling and the DerSimonian-Laird estimate for between-study variance were used in random effect model. When calculating the overall effect, the study showing a weight in random effect model below 5% was regarded as out-layer and not applied to analysis. A heterogeneity test to investigate common effect sizes and its dispersion among used studies was performed. Cochran’s Q test and the ratio of true heterogeneity to total variation in observed effects was represented by I² index in forest plot. Effect of CLA levels in diet was measured by meta-regression using the mixed-effect model. Statistical analysis and visualization of standardized mean differences were performed using the package ‘meta’ in the R statistical program (version 2.14.2) (R Development Core Team, 2010).

A total of 9 published research studies reporting the feeding trials of broiler chicken using dietary CLA were used as the data set and their experimental conditions were summarized in Table 1. The broiler breeds used were Cobb×Ross in one study, Arbor Acre in two studies and Ross breed in the remaining studies. CLA was incorporated into feed in different levels ranging from 0.25% to 5% and it was substituted with various oil sources in diet, completely or partially: soybean oil (2 studies), sunflower oil (2 studies), corn oil (2 studies), olive oil (1 study), beef tallow (1 study), hazelnut oil (1 study), flaxseed oil (1 study) and menhaden oil (1 study). Two studies did not mention substitution oils. Experimental days were varied from 21 to 42 d across studies. CLA was fed from d 1 of chick to the day of slaughter in 4 studies and the remaining studies started CLA feeding from mid to end of experiment resulting in an average duration of feeding of about 16.8 d. Major isomers in used CLA were c-9, t-11/c-12, t-10. Metabolizable energy in the diets ranged from 2,998 to 3,344 kcal/kg and crude protein was from 19.35 to 23.00%. The study of Badinga et al. (2003) was used for investigation of final body weight (FBW), and weight gain (WG). However, it was recognized as an outlier in both analyses because of low weight in the random effect model.

The overall effect of CLA feeding on final weight was calculated using 16 experiments in 6 studies (Table 2). Subgroups in the same study were designated according to used levels of CLA in diets or different CP levels in basal diet or different oils substituted by CLA. These different experiments in the same study were expressed as subgroups. For the calculation of the overall effect on WG, 16 subgroups in 7 studies were employed. Body weight gain (BWG) was measured by the difference of body weight between initial weight and FBW except for one study (Zhang et al., 2008) where average daily gain was applied. Feed intake (FI) was estimated from total consumed feed weight during experimental periods and 12 subgroups in 5 studies were used in calculation of the overall effect. The overall effect of CLA feeding on feed conversion ratio (FCR) was estimated by using 11 subgroups in 4 studies. The overall effect of CLA feeding on growth performance of broiler chicken is shown in Table 2. CLA feeding effect on FBW of broiler chicken was detected as negative in both fixed and random effects models. A significant negative effect means that the FBW at the treatment group (broiler fed CLA) was lower than the control group. A significant effect of treatment on FBW was found only in the fixed effect model (p<0.001). During the calculation of the overall effect, one study was rejected for its weight which was below 5% in the random effect model. In heterogeneity, the Q statistic was 361.61 and was significant (p<0.01) for FBW. With high and significant heterogeneity results, it could be suggested that the used studies or subgroups did not share a common effect size. Individual SMDs in subgroups and studies are shown in Figure 1A. The positive minimum value of 95% critical interval was found in four subgroups in the study of Aydin (2007), where CLA feeding was compared with flaxseed, menhaden oil and combination. The overall effect on WG was negative in both fixed and random models (p<0.01). The Q statistic and its probability on WG were 204.53 and p<0.01 respectively. Individual SMDs in all subgroups except studies no. wg9 to wg11 involved in the study from Zhang et al. (2008) showed negative effects (Figure 1B). Interestingly, where the study showed positive effects, CLA was fed from day one to slaughter. As for FI, a significantly negative overall effect of dietary CLA was detected in both fixed and random models (p<0.05). The Q statistic and its probability on FI were 100.27 and significant (p<0.01), respectively. The largely dispersed 95% confidence interval (CI) for each SMD over positive and negative from investigated studies were detected as shown in Figure 1C. The overall effect on FCR was significantly positive in both fixed and random models (p<0.05). The pattern of individual SMDs for FI (Figure 1C) show a converse pattern to that exhibited in Figures 1A and 1B.

The overall effects of CLA feeding on total SFA, USFA and C16:0, C18:0, C18:2 and C18:3 in thigh meat of broiler chicken are shown in Table 3. Total SFA content in thigh meat was significantly increased by CLA feeding in both of fixed and random effect models (p<0.01). The Q statistic was 134.53 and it was significant (p<0.01). In the calculation of overall effects, one subgroup involved in the study of Kim et al. (2008) was not used because of its low weight (under 5%). All subgroups applied to the analysis of total SFA were detected to represent a positive value for the lower limit of 95% CI in SMDs (Figure 2A). Significantly less total USFA content in thigh meat of the broiler fed CLA compared to control was found in both of fixed and random effect models (p<0.01). Three subgroups were declined for their low weight below 5%. The Q statistic was 87.64 and it was significant (p<0.01). Negative upper limits for the 95% confidence level were found in most subgroups except for four subgroups that showed relatively high weights (Figure 2B). Significant positive overall effects on C16:0 and C18:0 were found in both fixed and random effect models (p<0.01). Positive SMDs were found in all used subgroups (Figure 3A). The overall effect of CLA feeding on C18:3 was positive and the significant effect was found in only the fixed effect model (p<0.01). On the contrary, a significant negative overall effect was detected in C18:2 (p<0.01). The Q statistics for C18:3 and C18:2 were 43.12 and 193.87, respectively. Heterogeneities for the two fatty acids were significant (p<0.01). In forest plots representing individual SMDs, C18:3 (Figure 3C) and C18:2 (Figure 3D) showed quite different patterns from each other. The SMDs for C18:3 were highly dispersed (Figure 3C). Although there were large negative SMDs with small weight in C18:2, most SMDs with large weight were slightly negative (Figure 3D).

I² index values in all forest plots were relatively large. This suggests that there are possible effects from used subgroups. The levels of CLA in diets were used as one of the criteria in designing subgroups from studies. Therefore, a meta-regression analysis with different levels of CLA was then performed and summarized in Table 4. The coefficients for slope in all items in the growth performance category were negative, and significance was found only in the FBW (p<0.05). As for fatty acid profiles, a significant effect of CLA levels in diet was found only in stearic acid (p<0.01), and its effect was positive.

A lipid like substance derived from pan-fried hamburger supposed to contain anti-mutagenic function was firstly reported in 1979 (Pariza et al., 1979) and it was later identified as CLA. Since this finding, CLA has been researched in broad fields of science and reported to be positive to health. CLA is a collective term for geometric and positional isomers of linoleic acid, and all these isomers have a conjugated double bond instead of the methylene-separation (Schmid et al., 2006). The CLA content in foods derived from ruminant animal products is markedly higher than those from mono-gastric animals (Chin et al., 1992). The reason for particularly high CLA content in the food from ruminant products is explained by the action of rumen bacteria: biohydrogenation of linoleic acid - a phenomenon for the addition of hydrogen to unsaturated fatty acid (Badinga et al., 2003). Biological functions of CLA such as the suppressing activity against cancer, obesity and arteriosclerosis are well documented in many studies and these functions are regarded to be from its action on the modification of lipid metabolism and antioxidant activity (Ko et al., 2004; Hur et al., 2007; Zhang et al., 2008). Its health benefit activity was also suggested to be through the breakdown of endotoxin and reduction of prostaglandin secretion (Takahashi et al., 2002).

From the summary effects of CLA feeding on growth parameters in this study, it was found that CLA is not beneficial for growth performance of broiler chicken. This is consistent with the results from Simon et al. (2000), Du and Ahn (2003) and An et al. (2003). This negative effect of CLA was suggested to be arising from its function on the acceleration of fatty acid oxidation and consequently increased metabolic rate (West et al., 1998). However, CLA was also proposed to positively affect growth performance under stressful conditions such as endotoxin or vaccination through its antioxidant and reduction of inflammation activities (Zhang et al., 2008). These particular positive effects of CLA feeding under stressful situations were also found in the present work. There is one study (Aydin, 2007) where broiler chicks fed CLA under a stressful situation induced by vaccination showed positive SMD on FBW (Figure 1A). The positive effect of CLA feeding on growth performance under normal conditions can be found in other animals. CLA was reported to act as a fat-to-lean repartitioning in growing pigs and it can consequently result in an improvement in FCR (Dugan et al., 1997; Ostrowska et al., 1999). However, even though CLA can improve growth variables (weight gain and feed conversion ratio), it cannot be effectively positive when its incorporation rate is relatively high (exceeding 1.0%) (Szymczyk et al., 2001). However, it is not easy to find the result that shows positive effects of CLA feeding on growth performance particularly in broiler chicken. So, it might be concluded that CLA can obviously affect whole body metabolism of broiler chicken, although these do not translate to positive effects on growth performance. Therefore, feeding CLA with the aim of improving growth will not be suitable.

Many studies have proposed that CLA feeding can reduce fat deposition and increase lean meat content. These modifications might be explained by its activity on lipoprotein lipase to increase lipolysis (Park et al., 1997). Generally, many results for meat fatty acid profiles from broilers fed CLA show increased saturated fatty acids and decreased unsaturated fatty acids. These changes were supposed to be due to the inhibition of Δ⁹ desaturase activity in liver and consequent impaired conversion of C18:0 to C18:1 (Lee et al., 1998; Suksombat et al., 2007; Szymczyk et al., 2001). Another possible mechanism proposed was that CLA can be used as a substrate for Δ⁶ desaturase which converts linoleic acid to arachidonic acid resulting in a high content of linoleic acid in broiler chicken meat (Shin et al., 2011). However, in this study, the summary effect of CLA feeding on linoleic acid was negative and contradictory to previous findings. In this study, CLA feeding increased linolenic acid. This increase can be explained by differences in affinity between CLA and linolenic acid to Δ⁶ desaturase. CLA was reported to have a higher affinity to Δ⁶ desaturase than linolenic acid (Shin et al., 2011). This implies that dietary CLA is used for a substrate of Δ⁶ desaturase and subsequently it inhibits conversion of linolenic acid. Moreover, it was suggested that this increment of n-3 PUFA could be achieved by the incorporation of CLA from a particular oil source (e.g. flaxseed oil) (Du and Ahn, 2003). Even though, the effect of CLA feeding on its accumulation in meat was not investigated in the present study, the possibility of elevated CLA content in the meat of broiler was proposed and the supporting results were reported in a study by Sirri et al. (2003).