Metallomic profile in non-cirrhotic hepatocellular carcinoma supports a phenomenon of metal metabolism adaptation in tumor cells

We have previously described a form of hepatocellular carcinoma (HCC) in non-cirrhotic liver (HCC-NC) developed by Peruvian patients. We analyzed the metallomic profile in hepatic tissues from two independent cohorts exhibiting HCC-NC. Clinical, histopathological data, and HCC and non-tumoral liver (NTL) samples of 38 Peruvian and 38 French HCC-NC patients, were studied. Twelve metals were quantified using ICP/MS: Mn, Fe, Cu, Co, Zn, As, Se, Rb, Mo, Cd, Pb, and Sn. Associations between metals and survival were assessed. Our data showed significant differences between cohorts. Mean ages were 40.6 ± 20, 67.5 ± 9 years old for Peruvians and French, respectively. Fifty percent of the Peruvian patients were positive for the HBsAg, versus 3% in French patients. Mn, Cu, Zn, As, Se, Rb, Mo, Cd, Sn metal concentrations were higher in NTL of Peruvians. Importantly, metal concentrations were lower in HCC areas compared to NTL tissues in both cohorts, except for Cu for which mean concentration was higher in HCC (p < 0.05). Se concentration in HCC was associated with extended survival only in Peruvians. Our data, obtained in Peruvian and French HCC-NC cohorts, highlights similarity in the metallomic profile of HCC compared to NTL during the hepatic tumorigenesis in these specific groups of patients.


Introduction
HCC is the most frequent primary liver cancer, the sixth most common cancer and the second cause of death by cancer worldwide. 1 Many pathophysiological factors are potentially involved in the onset and progression of HCC, most of them being related to as chronic insult of the liver parenchyma. The development of HCC is commonly regarded a sequential multistep pathogenic process initiated with in ammation-mediated liver tissue damages and hepatocyte necrosis that induce liver brogenesis towards cirrhosis, which in turn increases the risk for HCC. 2 Thus, a larger number of HCC cases reported in the literature are found in cirrhotic patients, whereas HCC-NC are more rarely described.
Members of our group previously described a peculiar clinical and molecular presentation of HCC-NC developed by a signi cant fraction of patients in Peru. 3 These patients exhibited consistently remarkable clinical features: a) 50% of them are relatively young with a median age below 40 years old that includes children, teenagers, and young adults; b) the very large majority of the individuals presented with advanced-stage HCC and tumors larger than 10 cm-diameter; and c) Almost 80% of HCC occurred in noncirrhotic liver. 4 This peculiar clinical context where HCC-NC affects younger individuals was then corroborated to the whole region of South America. 5 We further substantiated this peculiar presentation of HCC-NC at both molecular and histological levels. First, Peruvian HCC displayed a unique mutation spectrum, in which the major class of alterations was epitomized by genetic short indels. 6 Second, liver parenchyma exhibited a very low level of in ammatory response and an absence of brotic process, and this, despite a strong prevalence of underlying infection with hepatitis B virus. 7,8 However, we observed within the non-tumor liver (NTL) parenchyma the presence of foci of cellular alteration in which cells are smaller compared to regular hepatocytes and exhibit an altered nuclear-cytoplasmic ratio. 7 These foci of cellular alteration showed also some degree of congruence with the co-expression of precancerous marker glutamine synthetase. Altogether, these ndings suggest that the clinical epidemiological situation encountered therein is due to some biological features intrinsic to the natural history of HCC in the population of Peru, and more broadly in South America. 9 This observation prompted us to search for additional pathophysiological cofactors associated with HCC-NC that could enhance the risk of developing at an early age among the Peruvian population.
It has been reported that the toxic effects of metals and their role as cofactor in occurrence of HCC. 10 For example, increased hepatic iron stores has been associated with HCC-NC. 11 Oxidative stress and reactive oxygen species promoted by siderosis are strongly suggested to be instrumental in HCC. Furthermore, As has been classi ed as an enhancer of oxidative stress and a human carcinogen, notably for HCC. 12 However, other metals such as Mn, Se, and Zn are required for normal activity of antioxidant defense system in cells in a concentration-dependent manner. 13 In order to evaluate the trace metal levels and whether metals are potential cofactor in the onset of earlyage HCC-NC in Peru, we performed a comprehensive analysis of metal concentrations in both HCC and NTL tissues of Peruvian patients (n = 38). In parallel, we contrasted these patients with another cohort of French individuals (n = 45) who developed HCC-NC in utterly different environmental, behavioral, and clinical contexts.

Results
Clinical data and histological data Table 1 summarizes the clinical and histopathological ndings between both cohorts. At time of diagnosis, the mean age of the Peruvian cohort was 40.6 ± 20.1 years-old. A large component of young patients was found in the Peruvian patient group. On the other hand, the mean of age for the French cohort was 67.9 ± 9.3 years old with a unimodal distribution (Fig. 1). A signi cant difference was found for the age of diagnosis between both cohorts (p < 0.001).

Determination of hepatic metal concentrations
Metal concentrations in NTL for both cohorts are reported in Table 2. Among 12 metals evaluated, nine of them including seven essentials (Co, Cu, Mn, Mo, Rb, Se, and Zn) and two toxic (As and Cd) had higher concentrations in NTL of Peruvian patients compared to French ones. Interestingly As was quanti able only in the Peruvian cohort. To the contrary, Sb concentrations were highest in French patient group compared to Peruvian group. Metals concentrations in HCC in both cohorts are presented in Table 3. Similarly to what was found in the NTL counterpart, the Peruvian patients had a higher concentration of metals in tumor tissues. As was found again exclusively in the Peruvian cohort, while Pb and Sn concentrations were higher in the French cohort. Concerning Fe, we must point that normal values of this metal are below normal iron concentration in hepatic tissue. Mean iron concentration values were three-fold higher in HCC of French patients (mean = 689.6 ± 1537.9 µg/g) than in Peruvian patients (mean = 255 ± 511.7 µg/g; p < 0.001).
Remarkably, these concentrations remain within the normal values for normal livers (Fig. 2). In both cohorts, we found that metal concentrations were higher in NTL that in HCC (Fig. 3) Based on the results of the Cox regression model, we performed an analysis to assess the impact of metals on survival. To this aim, we used the values of Se and Pb in HCC and NTL, to create two groups taking the mean concentration value as cut-off, above and below the mean. Our data shows that, the group of Peruvian patients with a Se concentration in tumor greater than 1.49 µg/g had a mean survival of 323 weeks versus 49 weeks for patients below this threshold (p = 0.033) (Fig. 4). Such result was not observed in the French cohort.

Discussion
Our study presents, for the rst time, a comparative analysis of liver metallomic pro les among patients from different geographical locations and genetic background, affected by HCC-NC, a disease that escapes the conventional pro le of HCC and representing merely 20% of cases of HCC. 14 However in the Peruvian context, HCC-NC represents 90% of cases of liver cancer. 3 The impact of metals on biological cell processes is a phenomenon not completely understood. Metals are involved in many bene cial functions like maintenance of pH, enzymatic cofactor, metabolism trigger, and reactive oxygen species formation as a product of normal metabolism. [15][16][17] However, while present in excess, metals may have harmful effects. The principal impact of metals is the disruption of intracellular redox balance, debt to increase of reactive oxigen species production. 18 Our data shows that concentrations of metals were higher in NTL of Peruvian patients compared to French individuals. Among those metals, some of them, such as As and Cd, are known to exert harmful effects on health. 19,20 Indeed, As and Cd are considered as carcinogenic to human according to WHO. The mechanism by which As contributes to the process of carcinogenesis is DNA damage with chromosomal aberrations, deletion mutations, and aneuploidy. 21,22 A strong link between exposure to arsenic and the development of HCC has been demonstrated in animal models, which also evidenced an increase in lipid peroxidation levels, prior to the onset of the brosis process and subsequent development of HCC. 21,23,24 The mechanism of Cd related injury involves the interaction and possible inactivation of thiol groups, leading to functional alteration of the metallo-enzymes of the superoxide dismutase family and to subsequent depletion of antioxidant agents such as glutathione. 25,26 We cannot assess whether HCC-NC and its spectrum of mutated genes are directly caused by the presence of heavy metals. 27 It is likely that metals could play an enhancing role in the carcinogenic process in association with hepatic carcinogenic agents, such the hepatitis B virus infection in Peruvian patients, 28 and/or alcohol intake that we were not able to assess with su cient con dence.
However, due to the absence of liver brosis in these series of patients, the role of ethanol is predicted to be marginal in the patho-physiological process.
Another important highlight is the common metallome pro le of HCC compared to NTL, with lower concentrations of metals in tumor tissue in both cohorts. Such ndings suggest that whatever the etiological factors, the geographic origin or other different parameters between cohorts, cancer cells have similar adaptive process regarding metal metabolisms. Whether the concentration decrease of most of these metals are related to a lower uptake, an increase release, and/or an increased turnover related especially to the enhanced cell cycle and cell metabolism remains not known.
The increase of Cu concentration, mostly in cancer tissues, was already reported in HCC, 10 and other cancers affecting especially breast, cervix, ovarian, and lung. 23,29,30 A hypothesis aiming to explain the behavior of Cu in cancer has been proposed by Fisher and collaborators, who states that the increase of this metal is due to a decrease in the catabolism in tumoral cells of ceruloplasmin (Cp) in tumor cells. 31 This phenomenon affecting the multicopper-carrying protein, might be due to increased sialylation produced by free sialic acid from cell membranes of neoplastic cells. This hypothesis was later supported in an animal model by Bernacki et al. 32 Another hypothesis concerns the role of copper as angiogenic agent. 33 McAuslan et al. showed that copper acts as promoter of endothelial cells migration. 34 Martin et al. consolidated both hypotheses, describing the link between Cu, ceruloplasmin and HIF-1α. 35 The authors showed that Cu acts as a stabilizer of the HIF-1α, mediating inhibition of prolyl-4-hydroylation.
The HIF-1α is eventually responsible for regulating transcription of many genes, including the Cp gene. Meanwhile, Himoto proposed a possible mechanism whereby Cu is required for binding HIF-1α to p300 and prevents the effect of Factor Inhibiting HIF-1 (FIH-1). 36 Another nding is the relationship between Se concentration in tumor tissues and survival in the Peruvian group. Se plays a major role in cell homeostasis, mainly through selenoproteins that are antiin ammatory, chemo-preventive, and immune modulators. 37 Cox regression model suggests a bene cial effect in overall survival for patients with higher levels of Se in HCC. This result is corroborated by a metaanalysis displaying the negative correlation between Se concentration and HCC. 38 Indeed, low levels of Se have been associated with high risk of developing HCC. Our model developed on a relatively small number of patients give us a signi cative outcome only in the Peruvian cohort. This could be explained by the fact that, in these patients, a "natural evolution of the disease" is observed, 4 since in this group of patients received only surgery, unlike the French patients who received additional local or systemic treatments that could modify HCC development. 39,40 Such ndings shall be corroborated in larger cohorts.
Finally, we must highlight the possible role of environment in the hepatocarcinogenic process due to the high mineral content of the subsoil and rivers. This statment was endorsed by several studies showing the presence of high heavy metal concentrations in the Andean regions of Peru. [41][42][43] In Egypt, Elwakil et al. have also described high concentrations of Cd, Pb, As, and Hg in blood samples from HCC patients who were exposed to the consumption of contaminated plants. 44 Therefore, we cannot rule out the relationship between the presence of metal in environment and natural history HCC.
Altogether, our ndings show that Peruvian and French cohorts of HCC patients have different metallomic pro les in NTLs, suggesting a putative impact of environmental and/or genetic factors. Whether these elements play a role in the very peculiar phenotype of HCC in Peru should be further explored. In addition, the modulation of metal concentration in HCC, that is shared by the two cohort suggests a coordinated modulation of metal metabolism in liver cancer cells during the carcinogenesis. Additional studies will allow to progress in understanding the role of metal metabolisms alterations in the hepatocarcinogenic process.

Ethical agreement
The present study investigated in strict accordance with the ethical principles contained in the Declaration of Helsinki.  48 .
The collision/reaction cell used was pressurized with a mixture of helium (93%) and hydrogen (7%); argon and hydrogen were provided by Messer ® . Ultra-pure water was provided from Milipore Direct-Q ® 3 water station. Nitric acid solution utilized at 69% (Fisher Chemical -Optima Grade ® ). The rhodium was used like internal standard (Fisher Scienti c ® ). Calibration ranges were realized using a multi-element solution (SCP Science ® Plasma Cal). The performance was calibrated using multi-element solutions, tune F and tune A (Thermo ® ). Certi ed reference material bovine liver ZC71001 was obtained from NCS Testing Technology (Beijing, China).

Statistics
Data collected were inputted into Numbers ® software version 5.3 (Apple Corporation). All statistical analyses were performed in R version 3.5.1, "Feather Spray" (R Foundation). The comparative analyzes between both cohorts were performed using the Mann-Whitney test. For survival analysis surgery date was considered as the starting parameter for the calculation of survival. Univariate Cox regression model was done with all variables 49 . Later all variables with signi cative values were used in a multivariate Cox model, the resulting model was improved with backward and forward stepwise regression. Finally, evaluation between all models was done with Akaike Information Criterion (AIC) index. Best model was selected from lowest AIC value. Finally, for each metal, we used the concentration average for create groups, i.e. over and under the average, and to develop comparative analysis of survival between these groups 6. Data availability All data generated or analyzed during this study are included in this published articule and its Supplementary information le.