Off-label prescribing of targeted therapies may occur in patients with other tumor types expressing the same target (referred as precision medicine)[7]. For instance, vemurafenib, a kinase inhibitor indicated in the treatment of metastatic melanoma with activating BRAF V600E mutation has been used off-label in refractory BRAF V600E mutation positive-hairy cell leukemia[8]. Furthermore, most of kinase inhibitors are not selective meaning that they display activity against other kinases not associated with approved indications[9]. For example, sorafenib is used off-label as a FMS-like receptor tyrosine kinase-3 inhibitor in relapsed acute myeloid leukemia[10]. Agents associated to phenotypic approaches, that is to say therapies against the tumoral vasculature (anti-angiogenic drugs such as bevacizumab) and new immunotherapies (checkpoint inhibitors such as ipilimumab) also carry the potential of alternative indications or combinations.

The prevalence of off-label use focusing on targeted therapies has not been investigated in detail. These studies are not easy to perform because these agents are numerous and the class is rapidly growing (more than 30 enzyme inhibitors approved worldwide since imatinib in 2001). In addition, they are both used in the in- and outpatient settings.

A swiss study has reported a low frequence of unsupported off-label use (7.8%) for 8 recent agents in a cohort of 985 consecutive patients under systemic anticancer treatment in 2012[11]. Variations were observed among these agents with almost no off-label use for pazopanib and a high level of unsupported use for bevacizumab (29.6%)[11]. The global prevalence (supported by the European Society of Medical Oncology and unsupported) was not reported for these 8 agents.

A similar study has been conducted in the United States in 2010 using patient database and focusing on the off-label use of some expensive intravenous agents (including the monoclonal antibodies cetuximab, rituximab, trastuzumab, bevacizumab)[12]. The frequence of off-label use was 30% of that half was clinically supported by the National Comprehensive Care Network (NCCN). Among agents, the rate of off-label utilization also varied considerably between trastuzumab (1%) and rituximab (67%)[12].

Another American study based on insurance administrative database found a rate of off-label use of rituximab of 25.3% during the period 2001-2007[13]. Around 50% of off-label use was evidence-based. Among targeted therapies, rituximab is the agent that probably carries the greatest potential for off-label indications mostly beyond oncology. Indeed, a Spanish prospective investigation reported that rituximab was the most frequently used agent off-label (21.1%) among 232 drugs (considering all therapeutic classes) in 5 tertiary hospitals during one year (2011-2012)[14]. In addition, a prospective Australian national study found that off-label use of rituximab covers 63 different diagnosis with 89% of off-label use outside oncology in the year 2012[15]. This is not surprising because rituximab is a non specific lymphocytical agent having potential numerous applications in the treatment of corticosteroid-refractory autoimmune diseases.

An Italian investigation described the off-label utilization of bevacizumab during the period 2006-2007 using patient database in the region of Lombardy[16]. The anti-angiogenic monoclonal antibody was mostly used (81.7%) in patients with metastatic colorectal cancer. On-label prescribing (according to the Italian Medicines Agency) represented only 241 (30%) of the 780 patients (i.e., first line treatment of metastatic colorectal cancer with fluorouracil-based chemotherapy). Off-label use concerned the timing of treatment in metastatic colorectal cancer (40%) and the use outside oncology in age-related macular degeneration (10%)[16].

More specifically, off-label use of anticancer agents has been investigated over a 10-years period (2000-2009) in a population of 2663 patients with breast cancer in the United States using an administrative data base[17]. A proportion of 13% of the patients were treated off-label mainly with cytotoxic agents. Regarding targeted therapies, off-label use of kinase inhibitors was anecdotal (0.4% of the patients). Off-label prescription of monoclonal antibodies was more prevalent (8% of the patients), particularly the agent anti-angiogenic bevacizumab (before the FDA cancelled the approval in breast cancer in 2011)[17].

Unsupported off-label use of the monoclonal antibodies panitumumab and bevacizumab has been retrospectively studied in a population of privatized insurance patients with metastatic colorectal cancer on progression, in the United States[18]. Between 2007 and 2010, off-label prescribing non-supported by the NCCN concerned 10% of the patients under bevacizumab and 16% of those under panitumumab.

Some studies have reported off-label use in populations of selected cancer patients with late-stage disease following tumor genomic testing. Preliminary experiences have described the opportunity of using tumor genomic information to guide a specific treatment in certain patients through so-called molecular tumor boards[19]. In a prospective study including 250 adult patients mostly with colorectal, breast, lung and pancreas cancers, only 10% of the patients tested could be treated mainly through a clinical trial[20]. Overall, following tumor profiling, off-label use only represented 2.8% of the patients[20]. Le Tourneau et al[21] investigated molecular tumor profiling representing 3 pathways (hormone receptor, PI3K/AKT/mTOR and RAF/MEK) in 741 patients with refractory metastatic disease. Around 40% of the patients were eligible to a panel of 11 off-label targeted therapies including imatinib, dasatinib, vemurafenib, sorafenib, erlotinib, lapatinib, trastuzumab and everolimus. Currently, these new strategies of treatment are not expected to bring extended off-label use because a minority of genomic alterations (10%-40%) are targetable or “druggable”. Furthermore, the delay of treatment may impede access for patients whose disease progresses and in some countries like the United States, patients could be denied from treatment for covering reasons.

Overall and based on these preliminary data, off-label use of targeted therapies appears to be in the same range as that regarding older drugs (6.7%-33%) with wide variations among agents[1].

As seen above, Le Tourneau et al[21] evaluated in a randomized phase 2 trial the clinical impact of selected molecular tumor profiling. Unfortunately, among the treated and randomized patients with “druggable tumors” (n = 195), the off-label use of targeted therapies did not improve the progression free survival (primary endpoint) when compared with those treated by chemotherapy according to the oncologist choice (around 2 mo in both arms).

A French registry has collected the off-label use of kinase inhibitors in 249 patients with sarcomas mainly pretreated (89%)[22]. Sarcoma is a very heterogeneous disease with little therapeutic options. Decision of off-label treatment was made following discussion with experts, based on a scientific rationale (96%). Sorafenib (45%), sunitinib (25%), sirolimus (9%) and imatinib (8%) were mostly used. Toxicities above or equal to grade 3 were observed in 32% of the patients. The median progression-free survival was 4.1 mo (Interval of confidence or 95%CI: 3.2-4.8) and overall, the results were judged similar to those of published trials[22].

In 2010, off-label use of the multi-kinase inhibitors sunitinib and sorafenib has been reported in 15 patients with follicular/papillary radioactive iodine refractory cancer[23]. The progression free survival was 19 mo. Since then, sorafenib gained its approval in 2013 based on a phase 3 randomized study which showed a significant improvement in the progression free survival (10.8 mo vs 5.8 mo with placebo)[24].

Clinically, the limits of off-label use are a lack of activity and/or the appearance of serious side effects. Tumor types carrying the same mutation and potentially eligible to off-label treatment do not respond uniformly to the targeted therapy as it has been shown in cancers with BRAF V600 mutations[25]. Furthermore, some off-label combinations may be detrimental. Ipilimumab and vemurafenib are both targeted therapies used as a single agent-treatment in patients with metastatic melanoma. The trial evaluating the association has been closed for safety reasons (hepatotoxicity) underscoring the risk of co-administering off-label recent agents with new mechanisms of action[26].

Regarding the re-birth of immunotherapy, this phenotypic approach that induces a T-cell response against tumour is susceptible to be used in any kind of cancer. However, the reality is more complex and the responses vary among patients and cancers. So, the exploration of new indications should be devoted to clinical trials. Otherwise, even if in case of positive and promising results (associations of immunotherapies in metastatic melanoma), enthusiasm should be tempered[27]. Indeed, ipilimumab with nivolumab has been shown to be superior in terms of median progression free survival than the immunotherapies given alone (11.5 mo vs 6.9 mo for nivolumab and 2.9 mo for ipilimumab). The association was also more toxic[27]. Ipilimumab and nivolumab are currently approved as a single agent-therapy in metastatic melanoma. The off-label use of combinations of these checkpoint inhibitors is premature and is not sustainable for financial reasons.

Due to their astronomical pricing, covering recent anticancer agents in their labeled indication is a major concern in most health systems. So, off-label prescribing adds reimbursement difficulties. Covering of off-label use depends on the country, the level of clinical evidence and can constitute a barrier for certain patients. Loss of patent of monoclonal antibodies and kinase inhibitors and the forthcoming arrival of less costly biosimilars (rituximab) as well as generics may improve access in well supported clinical situations.