The diagnosis of cirrhosis was made on the basis of:

- Clinical arguments including ascites and edema; morphological changes in liver, portal hypertension and hepatocellular insufficiency;

- Biological arguments with hepatocellular insufficiency (in particular prothrom- bin rate less than 60%, albuminemia less than 35 g/liter, beta-gamma block in the protidogram);

- And/or echographic and possibly endoscopic arguments.

Anemia was defined when the hemoglobin rate value was below 12 g/dl. We were able to characterize anemia using erythrocyte constants:

- A hypochromic microcytic anemia for a Mean Corpuscular Volume (MCV) value < 80 femtoliters and the Mean Corpuscular Hemoglobin Content (MCHC) < to 27 picograms;

- A macrocytic anemia for a MCV value greater than 100 femtoliters;

- A normochromic normocytic anemia for a MCV value between 80 and 100 femtoliters and MCHC between 27 and 31 picograms.

Descriptive statistics were performed and the results were presented in the form of effective and proportion tables for the categorical variables. Quantitative variables were presented as means with their standard deviation and extreme values. Qualitative variables were compared using the chi-square and/or Fisher test. Analyzes were carried out using R version 3.3.2 software. The significance level was set at 5%.

During the study period, we collected 253 cases of cirrhosis, of which 153 patients had anemia (60.5%); there was a male predominance (73.2%) with an M/F sex ratio of 2.64. The mean age was 51 ± 13 years with extremes of 20 to 99 years. The B viral origin of cirrhosis was the most common (60.1%) (Table 1).

The main complication was oedemato-ascitic decompensation (82.4%) followed by hepatocellular carcinoma (34%). The Child-Pugh B score was the most repre- sented (74.5%). Complication data, Child-Pugh score are noted in Table 1. Of the 33 patients with hepatic encephalopathy, 15 (45.5%) had hepatocellular carcinoma.

Hypochromic microcytic anemia was noted (48.4%) followed by normochromic normocytic anemia (46.4%) as shown in Table 1. The mean values of the various parameters of the complete blood count are noted in Table 2; 82 patients (53.6%) had thrombocytopenia; 12 patients (7.8%) had neutropenia; pancytopenia was noted in 17 patients (11.1%).

Macrocytic anemia has been noted in men than in women (p = 0.486) and normochromic normocytic anemia has been noted in patients over 50 years of age (p = 0.664) as shown in Table 3. Hepatitis B virus was most commonly found with 50% hypochromic microcytic anemia followed by 46.7% normochromic normocytic anemia (p = 0.311) as shown in Table 3. The ethylic cause was noted in 50% of patients with macrocytic anemia (p = 0.311).

Hepatic encephalopathy was significantly more frequent in patients with hypochromic microcytic anemia (45.5%) (p = 0.025); hepatocellular carcinoma was significantly noted with 63.5% hypochromic microcytic anemia (p = 0.016) as shown in Table 3.

Child-Pugh C score with 47.4% hypochromic microcytic anemia was more frequent (p = 0.673).

No case of hypochromic microcytic anemia in our study performed workup to evaluate iron metabolism for iron deficiency. In our study, this was the first type of anemia found in cirrhotic patients. Its frequency in our study is higher than that of Nacoulma et al. [3] (20%), Denie et al. [2] (38%). Even minimal and chronic digestive haemorrhages, chronic inflammation, hypotransferrinemia [9] [10] by limitation of transferrin synthesis due to hepatocellular insufficiency could explain the microcytic hypochromic anemia during cirrhosis.

It would have been useful in our study to have the reticulocyte count in order to judge the regenerative or non-regenerative nature of normochromic normocytic anemia. Digestive hemorrhage, hypersplenism frequently encountered in cirrhosis, and hemolysis [11] could explain the normochromic normocytic anemia encountered in cirrhosis. Bone marrow dysfunction due to hepatitis viruses has also been suggested as a cause of aplastic anemia [12]. In our series, hepatitis B and C viruses are the main causes of cirrhosis and nearly 75% of patients with normochromic normocytic anemia had a B or C viral cause. The frequency of normochromic normocytic anemia in our study was higher than that found in the studies by Nacoulma et al. [3] (43.3%), Denie et al. [2] (12%). In our study 51.7% of patients with normochromic normocytic anemia had esophageal varices; however, none of them had digestive hemorrhage during hospitalization.

The low frequency of macrocytic anemia in our study is probably related to the etiology of cirrhosis which is dominated by hepatitis B and C viruses. This frequency is lower than that found by Dénie et al. in France (48%) [2]; this could be explained by the fact that the causes of cirrhosis in France are dominated by alcohol [13]. Macrocytic anemias could be explained on the one hand by vitamin deficiencies, in particular folates and vitamin B12 relating to their uptake and storage [14]; data on these vitamin deficiencies were not available in our patients’ medical records; sometimes the financial difficulties of the patients explain the non-rea- lization of these dosages; however, it seems clear that their dosage is necessary for the care of cirrhotic patients who are at risk of malnutrition and on the other hand by chronic ethylism which frequently leads to macrocytosis. In our study, alcohol was noted in 50% of cases of macrocytosis. This explanation is confirmed by Belaiche et al. [15] for whom, blood macrocytosis in cirrhosis is linked not only to the toxic effect of alcohol on hematopoiesis but also to folate deficiency. The bone marrow toxicity of alcohol and its metabolite acetaldehyde is responsible for bone marrow failure in alcoholics [1].

In our study, more than half of the patients had thrombocytopenia. Thrombocytopenia was found in 66.67% of cirrhotic patients significantly in the study by Sack et al. [4] and people suffering from cirrhosis and/or hepatocellular carcinoma are 43 times more likely to develop thrombocytopenia than controls (Odds ratio = 42.66) [4]. In the study by Nacoulma et al. [3], thrombocytopenia was found in 59.50% of cirrhotics. Pancytopenia was noted in our study in 11.1% of cases; hypersplenism is one of the identified causes of pancytopenia [16]. In our study, 20.9% of patients had splenomegaly. The mechanism of this pancytopenia could be either a peripheral destruction of the various blood lines or a lack of bone marrow production. Sack et al. [4] in their study noted erythrocyte morphological abnormalities such as anisocytosis, erythrocyte poikilocytosis, presence of stomatocytes, target red blood cells. Our study did not do so because data regarding these erythrocyte abnormalities were not available in the patients' medical records.

Hepatic encephalopathy was significantly more frequent in patients with hypochromic microcytic anemia; this could be explained by the fact that in our study, all cases of digestive hemorrhage (gastrointestinal bleeding) had hypochromic microcytic anemia and digestive hemorrhage is well known to be one of the causes of hepatic encephalopathy in cirrhotic patients; this alone could not explain this significant link as there were only 4 cases of gastrointestinal bleeding in our sample. The Child-Pugh C score with 47.4% hypochromic microcytic anemia was more frequent but not significantly. Hepatic encephalopathy is one of the parameters of the Child-Pugh score; this could also explain the fact that hepatic encephalopathy is found among patients with microcytic hypochromic anemia because the latter have a poor prognosis Child-Pugh C score. Hepatocellular carcinoma was significantly noted with 63.5% hypochromic microcytic anemia as well as hepatic encephalopathy was significantly noted with hypochromic microcytic anemia; this correlates with the fact that of the 33 patients with hepatic encephalopathy, 15 (45.5%) had hepatocellular carcinoma.