Prognostic factors of survival in HIV/HCV co-infected patients with hepatocellular carcinoma: the CARCINOVIC Cohort

Background and Aims : HIV/HCV co-infected patients with hepatocellular carcinoma (HCC) have poorer survival than HCV mono-infected patients. We aimed to evaluate the prognostic factors for survival. Methods : From 2006 to 2013, 55 incident HCCs among HIV+/HCV+ patients, from three ANRS cohorts, were compared with 181 HCCs in HIV-/HCV+ . palliative treatments were decided in HIV+/HCV+ patients (51% vs. 19%, P <0.001). The 1 and 2-year crude survival rates were 61% vs. 78% and 47% vs. 63%, P =0.003, respectively. In a Cox model multivariate analysis adjusted for the cohort, age and sex, the most important prognostic factor for survival was the infiltrative form of the tumour (aRR: 8.10 [4.17-15.75], P< 0.001). Conclusions : The radiological aggressiveness of the tumour is the best prognostic factor associated with poorer survival of HCC in HIV+/HCV+ patients. High α-foetoprotein level and decompensated cirrhosis are other ones. This justifies a particular attention to the detection and the management of small nodules in this high-risk population.


Introduction
Hepatocellular carcinoma (HCC), a primary liver cancer, is the third leading cause of cancer deaths worldwide and the sixth most common malignancy (1). Before new oral hepatitis C virus (HCV) agents, the frequency of HCC in human immunodeficiency virus (HIV)-infected patients has increased worldwide in parallel with improvements of HIV treatment regimens and HCV co-infection, estimated in 16% to 33% of HIVinfected patients (2,3). Recently, the D:A:D study, which followed 49,000 HIV positive persons in Europe, United States and Australia calculated an incidence rate of end-Accepted Article stage liver disease and HCC of 1 per 1,000 person-years, including a third of HCC (4).
In the Era of Direct-Acting Antiviral Agents, there is no data concerning the incidence of HCC in HIV/HCV co-infected (HIV+/HCV+) patients but it is likely to fall in this population, in line with its predicted decline in HCV mono-infected (HIV-/HCV+) patients (5,6).
Nevertheless, the clinical course of HCC in an HIV-infected setting is still not clearly defined; most studies have involved small samples and many of the HIV patients included were not receiving combined antiretroviral therapy (cART). Two case-control studies reported that HCC occurred at a younger age, with more infiltrating and metastatic tumours in HIV+/HCV+ than HIV-/HCV+ patients (7,8). Moreover, elevated α-foetoprotein (AFP) levels were quite common in the HIV+/HCV+ group, independently of the HCC staging score and the severity of cirrhosis (8,9). We recently demonstrated that HCC present with more severe radiological features at diagnosis in HIV+/HCV+ than HIV-/HCV+ cirrhotic patients because of the frequent infiltrative types HCC with portal vein tumour thrombosis (11). The overall survival of patients who develop HCC is also significantly worsened in HIV+/HCV+ than HIV-/HCV+ patients (7)(8)(9)(10)(11). However, the therapeutic management and the analysis of prognostic factors for survival in prospective studies have not been taken into account to date. We therefore constructed a large Carcinovic collaboration from three ANRS cohorts (Prethevic, HepaVih, CirVir), to include all HIV+/HCV+ patients diagnosed with HCC in order to analyze the prognostic factors for survival and to evaluate their therapeutic management.
Cases of HCC diagnosed at imaging in HIV+/HCV+ patients until January 2013 were prospectively included in the Carcinovic collaboration. A control group of HIV-/HCV+ cirrhotic patients with HCC was included from ANRS CO12-Cirvir.

Diagnosis of HCC
All patients included in this study had cirrhosis, the diagnosis was established histologically or by transient elastometry (Fibroscan treshold value 12 kPA). In the event of a focal nodule identified by US examination, with or without elevated AFP levels, the diagnosis of HCC was confirmed on imaging according to the updated 2011 American Association for the Study of Liver Disease (AASLD) criteria (15): the presence of a nodule larger than 1 cm in diameter with arterial phase hyperenhancement and portal venous or delayed phase washout at either computed tomography (CT) or magnetic resonance imaging (MRI) (15). Image analysis was performed unblinded on a picture archiving and communication system station jointly by two expert abdominal radiologists (ML and OS) in the Carcinovic collaboration.
Image analysis evaluated the type of HCC (nodular or infiltrative), the nodule's largest diameter and the presence/absence of a tumour portal obstruction, defined as a distension of the portal vein lumen by the enhancement of a thrombus (9). Reports of imaging techniques showing liver focal lesions were secondarily reviewed by two senior hepatologists (MGS and PN).
All patients were evaluated from the date of first vascular imaging with radiological evidence of HCC, from which started the follow-up.

Virological analysis
HCV and HIV positivity was identified by the presence of serum HCV and HIV antibodies, plus HCV and HIV ribonucleic acid, respectively. The first positive serological tests were considered as the initial dates of viral infection.

Data collection
Follow-up was ensured every 3 months until death, loss to follow-up, or liver transplantation (LT) in the Carcinovic collaboration. At each visit, a physical Accepted Article examination, Model for End-Stage liver Disease (MELD) score, Child-Turcotte-Pugh (CTP) score, CDC score, and virological, immunological, biochemical and drug toxicity assessments were carried out. The treatment histories and patient's responses to previous treatments of HIV and HCV infections were recorded at the diagnosis of HCC and during follow-up.
Twenty-six Infectious Diseases or Hepatology units in France participated in the Carcinovic collaboration. A common database with, patients' demographics, radiological and biological features, modality of therapy and survival, was created, which represented the support for analysis.
Data on HIV-/HCV+ patients from ANRS CO12-Cirvir were extracted and integrated into the whole database for the analysis.

Therapeutic management of HCC
The treatment of HCC, defined according to the Barcelona Clinic Liver Cancer (BCLC) staging system was recorded at initiation and at each modification during follow-up (16). Liver transplantation, percutaneous ablation (PA) and surgical resection were considered as curative treatments. Percutaneous ablation included radiofrequency ablation or alcoholization. Transarterial chemoembolization (TACE) was also considered as a curative treatment when associated with PA or surgical resection.
First, the eligibility to LT was evaluated through the application of Milan criteria at diagnosis of HCC; retrospectively compared with secondly the AFP score (17,18). The access to LT was completed at each visit by a questionnaire about the indications, contraindications, enrolment on waiting list and LT.
Palliative treatments consisted of TACE, chemotherapy or anti-angiogenic therapy given alone or in combination.

Statistical Analysis
The characteristics and outcomes of HCC were analyzed, as were the therapeutic strategies employed, survival rates, prognostic factors for survival and the causes of death. Kaplan-Meier methods were used to calculate the 1 and 2-year survival rates since the first confirmed diagnosis of HCC by MRI or CT. Survival was studied from the date of confirmed diagnosis until death, censoring or LT. The censoring date for patients who were neither deceased nor transplanted was the date of the last visit until

Accepted Article
. January 2014. The relationships between overall survival and the following parameters were studied: gender, age at HCC diagnosis, AFP level, HIV and HCV viral loads, diabetes mellitus, type (infiltrative or nodular) of tumour, venous portal thrombosis, alcohol and tobacco consumption. Univariate and multivariate Cox regression models were used to assess prognostic factors for survival. All analyses were performed using SAS v9.3 software (SAS Institute, Inc, Cary, NC). A P value <0.05 was considered to denote statistical significance.

Ethical aspects
The three cohorts were implemented according to the Declaration of Helsinki and the French law relative to biomedical research, with written informed consent obtained from all patients. The three cohorts were approved by their relevant Ethics Committee (CPP Ile-de-France) and the French Agency for the Safety of Medicines and Healthcare Products (ANSM).
They were compared with 181 cases of HCC developed in HIV-/HCV+ patients with histologically proven cirrhosis from the ANRS CO12-CirVir cohort.
The relevant characteristics concerning these populations are shown in Table 1 The main mode of HIV/HCV acquisition was intravenous drug use (87%), which was associated with alcohol intake in 76% of cases and tobacco intake in 89%. HCV genotype 1 was the most frequent in both groups (55% vs. 80% in HIV+/HCV+ and HIV-/HCV+, P<0.001, respectively), followed by genotype 4 (22%) in HIV+/HCV+, and 3 (12%) in HIV-/HCV+ subjects.
There was no difference in AFP levels at the time of diagnosis between HIV+/HCV+ and HIV-/HCV+ patients, with a majority of low AFP levels below 200 ng/mL (Table 1).

Radiological characteristics of HCC at presentation in the two cohorts
All cirrhotic patients were prospectively followed at least every 6-months as currently recommended. In the HIV+/HCV+ patients, HCC was mostly diagnosed as a result of this regular follow-up (78% in HIV+/HCV+ cohort and 100% in HIV-/HCV+ cohort); except for 5 patients (9%), thanks to incidental imaging in additional to the screening programme, and for 7 patients (13%), in the context of suggestive clinical symptoms.
The median delay between the first suspicion of a hepatic nodule by US (suspected date of HCC) and the vascular imaging that confirmed the diagnosis of HCC according to AASLD radiological criteria (diagnostic date of HCC) was 0.56 months [0-1.81]. In 49 (89%) patients, this period (defined as the diagnostic interval) was 3 months or less, while in 6 (11%) patients, it was longer than 6 months (with a maximum of 15 months). Long diagnostic interval > 6 months was due to an atypical nodule on vascular imaging that did not comply with AASLD criteria in four patients, while in two other patients there was a lack of compliance. Vascular imaging revealed an infiltrative HCC in 24% of HIV+/HCV+ patients vs. 14% of HIV-/HCV+ patients, P<0.001, respectively, and a nodular form in 76% vs. 96% of cases, P<0.001, respectively, which largest diameter was 23.5 mm [17][18][19][20][21][22][23][24][25][26][27][28][29][30] in HIV+/HCV+ cohort and 20  mm in HIV-/HCV+ cohort. The 13 HIV+/HCV+ Accepted Article patients with infiltrative forms also suffered from a tumour portal thrombosis, which accounted for 81% of all tumour portal thrombosis observed.
Therapeutic management of HCC in the two cohorts and access to liver transplantation in HIV+/HCV+ patients As reported in Table 1, overall by the end of follow-up, HIV+/HCV+ patients had received less curative treatment than HIV-/HCV+ ones (38% vs. 61%, P<0.001, respectively) however, LT tended to be more frequent in HIV+/HCV+ than HIV-/HCV+ patients in 9 (16%) vs. 11 (6%), respectively.
The access to LT was evaluated in 39 (71%) HIV+/HCV+ patients without an infiltrative form or tumour portal thrombosis (Fig. 1). Finally, 33 (60%) and 36 (65%) patients, respectively met the Milan criteria and AFP score at diagnosis of HCC. For patients in Milan criteria, LT was finally contraindicated in 17 (31%) patients: this was due to a tumour extension during follow-up in 5 patients (4 with the development of new hepatic nodules and 1 with haemorrhagic ascites) leading to a fatal outcome, severe alcoholism in 7 patients, advanced age in 1 (74 years old), lymphoma in 2 and psychiatric disease in 2. Seven of these 17 patients received an alternative curative treatment. Among the 16 (29%) patients eligible for LT, three died before LT from liver failure or bacterial septicaemia. As the latter patients had received TACE during the waiting period, they belonged to the palliative group at the end of follow-up. Among the 13 listed patients, four were treated with alternative curative treatments without recurrence of HCC and didn't need LT; only nine underwent LT, while still complying with the Milan criteria and AFP score. Six (67%) of the latter, required pre-operative treatment (PA in five or TACE + PA in one) during the waiting period.

Causes of death and prognostic factors
In the Carcinovic collaboration, the median follow-up period since the diagnostic date of HCC was 11.96 months [5.51-27] HIV+/HCV+ patients with HCC had worse survival than HIV-/HCV+ patients, (Fig 2). adjusted for the cohort, age and gender, was independently associated with the probability of death (Table 2).

Discussion
This is the largest multicentre prospective cohort of HIV/HCV co-infected patients with HCC to have been studied. We were able to confirm that the latter have poor survival (61% and 47% at 1 and 2 years) and poor access to curative therapies (38%), compared to HCC in HIV-/HCV+ patients. Trinchet et al. reported that curative treatments were implemented in 71.3% of HIV-/HCV+ patients, with PA in 50% of cases (14). Moreover, in their study, only 9.1% of patients (n=9) were transplanted, although 78% met the Milan criteria, as the data were given in intent to treat (14).
Supporting our data, Merchante et al. found that only 30% of their HIV+/HCV+ cohort met the Milan criteria for LT at diagnosis of HCC; finally, 13% received potentially curative treatments (19). In addition, these two studies highlighted that a minority of HCC in HIV+/HCV+ patients achieve curative treatments unless they are potentially curable at diagnosis (14,19). In fact, in our cohort, where 60% of patients met the Milan criteria, nearly half of them were not eligible for LT because of extra-hepatic Accepted Article contraindications or death (10%) while on the waiting list from hepatic causes (mainly liver failure). Another curative treatment than LT was implemented in 10% and nearly 30% of the patients falling within the Milan criteria were transplanted. Currently, these data support the idea that the diagnosis of HCC in HIV+/HCV+ patients is globally performed in too advanced stages of cirrhosis and HCC, which lead to a poor access to curative treatment.
First, the poor prognosis for HCC in HIV+/HCV+ patients may be explained by a late diagnosis. However, our results, which suggest that the diagnostic interval, i.e., between the date of suspicion of a nodule and its confirmation, is not particularly long (median diagnostic interval: 0.56 months [0-1.81] in reference centres and a 6-months follow-up interval), do not support this hypothesis. Surprisingly, the individuals with longer diagnostic interval had not necessarily poorer survival. Unfortunately, we have no data about the diagnostic interval in the mono-infected Cirvir cohort, prospectively followed by a physical exam and US every 6 months.
Second, the problem should be a more rapid progression of HCC in HIV+/HCV+ patients. Indeed, our group previously reported a more advanced radiological presentation at diagnosis in HIV+/HCV+ than in HIV-/HCV+ cirrhotic patients, related to more infiltrative tumours (23%) and tumour portal thrombosis (28.5%) (11). Here, on the basis of a larger sample, the infiltrative form of HCC, which affected 24% of HIV+/HCV+ patients, was a strong prognostic factor for death (adjusted RR of 8.10  (20). In HIV patients, there is also an alteration in the balance between CD4 and CD8 T-cell activities, with a predominant CD8-cell response mediated by cytokines such as IL-4, IL-5, and TGF-β, which increases liver inflammation and fibrosis (20). It appears that HIV proteins (tat, gp120), or HIV itself, promotes pro-inflammatory cytokines by binding (and possibly entering) hepatic stellate cells on its CCR5 receptors, which is also involved in hepatocarcinogenesis (21,22). Finally, there is the role of metabolic syndrome associated with HIV infection.

Accepted Article
These data therefore argue for an earlier diagnosis of HCC in HIV+/HCV+ patients, and a new screening policy could be proposed because of the rapidly evolving form of these tumours. Current international guidelines that recommend US screening every 6 months in cirrhotic patients seem to be inappropriate in the latter patients. Some authors proposed a 3-month interval for US screening in HIV/HCV patients, rather than 6 months. However, no prospective studies have validated such a strategy and European guidelines recommended 6-monthly liver US and AFP testing in HIV+/HCV+ patients with cirrhosis (23). Admittedly, the sensitivity of the US is too limited to detect both infiltrative and nodular types of HCCs, MRI is frequently required to confirm the diagnosis. Indeed, MRI or CT scan have been proposed for screening HCC in particularly high-risk population in Asia (24).
In cases where MRI fails to conclude, biopsy remains the last resort. Obtaining an earlier HCC diagnosis in HIV+/HCV+ patients should improve access to, and the efficacy of, curative treatments. This is especially true regarding the treatment of infiltrative type HCC, which is limited to palliative approaches (25).
As well as radiological prognostic factors, immunosuppression, well known to accelerate tumour progression, has been confirmed as constituting a higher risk of HCC (10), which was not supported by our study.
We acknowledge that one limitation of our study is the lack of a prospective evaluation of radiological tumour progression in order to give more precision in the prognostic imaging factors.

Accepted Article
For qualitative values, results are expressed in n (%), for quantitative values, results are expressed in median with range.