The loss of a limb after amputation, altering motor and sensory functions, severely affects the daily lives of patients (Pasquina et al., 2006). Today's Robotics is able to offer sensorized mechatronic hands with amazing performance; to these devices, conventional body-powered or electromyographic control systems revealed to be inadequate. In order to overcome such issue, novel advanced neural prosthetic devices, which attempt to interface bidirectionally with the user through invasive microelectrodes implanted inside stump's peripheral nerves (Di Pino et al., 2013), are in developmental phase (Di Pino et al., 2009). Basically, these neural interfaces are used to deliver afferent information to the nervous system (sensory feedback) (Benvenuto et al., 2010; Di Pino et al., 2012; Raspopovic et al., 2014), and at the same time, to extract consciously-modulated neuronal signals (motor volition) from healthy portions of the nervous system (Figure 1). Then, through a variety of signal processing algorithms, these signals can be used to drive an external device, such as a prosthesis (Rossini et al., 2010). Such interfaces have the potential to bring crucial advancements to the user: (a) to get a more complex signal, compared to the muscular one, allowing to control more degrees of freedom of the prosthesis; (b) to transmit the sensory information to the central nervous system (CNS) allowing to feel and experience the environment; (c) to restore the original closed “sensory-motor loop” control, meaning to have prosthesis and user linked by a double connection where information can travel from one to the other in both directions. A closed loop control allows to accurately exploit in real time sensory information to drive motor command, while having the feed-back of the outcome of the command and constantly readjust it to reduce any deviation from the desired targeted action.

When developing new neural interfacing devices, the attempt of the researchers is, on one hand, to preserve some important features, such as biocompatibility and selectivity properties, and on the other hand, to minimize the correlated potential counterpart, such as immune response reaction, the invasiveness and the risks of the implanting procedure. In this review, we focus our attention on biological challenges, mainly because all the efforts to develop a more immune system-friendly recording/stimulating device will also prevent the complete loss of functionality over time and, in turn, the loss of all the improvements done for reaching high selectivity and implantable ease.

The placement of any therapeutic biomaterial-based device in the in vivo environment requires surgical actions that damage the target tissues (Anderson, 1988). Following those damages, all the processes to restore tissue homeostasis around the implant are part of the physiological regeneration called wound healing (Teller and White, 2009).

However, the continuous presence of any medical implants contributes to a long-term overstimulation of the immune system, which leads to chronic inflammation and poor wound healing. This unbalanced reaction, also known as the foreign body response or reaction (FBR), determines the implant failure and might have contributed to implant reduction functionality in human (Dhillon et al., 2005; Rossini et al., 2010; Horch et al., 2011; Raspopovic et al., 2014) and animal trials (Lago et al., 2007b; Badia et al., 2011; Groothuis et al., 2014; Harreby et al., 2015; Wurth et al., 2017), due to the formation of a cellular, protein mediated, capsule around the implant.

While it has been suggested that encapsulation is beneficial in interfaces, such as epimysial and cuff electrodes as it stabilizes the implant (Grandjean and Mortimer, 1986; Grill and Mortimer, 1998), FBR is an important issue to be solved in intraneural electrodes (Navarro et al., 2005), limiting device function over time, often forcing a premature removal (Badia et al., 2011). We can divide the FBR process in four major steps: (1) blood-plasma proteins adsorption to the foreign body; (2) monocyte recruitment and differentiation to macrophages; (3) macrophages activation and fusion to form giant cells; (4) fibroblasts recruitment and activation to form fibrotic tissue (Figure 2). Regardless of the tissue or organ into which a biomaterial is implanted, the initial inflammatory response is activated by injury to vascularized connective tissue. Immediately after implantation (step 1), proteins including fibronectin, vitronectin, albumin, fibrinogen and complement factors derived from blood-plasma will adsorb to biomaterial surfaces, generating a blood-based barrier that causes thrombosis, guides the movement of monocytes and neutrophils to implant surface and activates complement and coagulation cascade (Szaba and Smiley, 2002). Monocyte recruitment and differentiation to macrophages (microglia in CNS) (step 2) may continue up to weeks, and chemotactic factors are released over longer periods of time (Tang and Eaton, 1999). Plasma fibronectin, among all plasma proteins, after coating the implant surface, changes its conformation and seems to act specifically as receptor for fibroblast and additional macrophages (Keselowsky et al., 2007). This persistent inflammatory stimulus leads to chronic inflammation confined to implant site. More in details, the cytokine release (e.g., IL-4 and IL-13; IL-1β, and TNFα in CNS) from the microenvironment (Shen et al., 2004) leads to the fusion of the accumulated macrophages to form foreign body giant cells (FBGC) (step 3) (Anderson, 2000). Original purpose of FBGC formation, producing catabolic enzymes and acids, is to try to digest the foreign body; its side effect is the possibility of causing implant damage. Macrophages are probably the most important cells in chronic inflammation, because of the great number of biologically active products they produce, including multiple growth factors that are extremely important to promote: (i) further recruitment of monocytes/macrophages cell population; (ii) growth of fibroblasts and their differentiation in myofibroblasts (astrogliosis in CNS); (iii) blood vessels and epithelial cells regeneration (Anderson et al., 2008). The proliferation of myofibroblasts (step 4) in developing granulation tissue leads to active synthesis of extracellular matrix (ECM) components, such as collagen and proteoglycans to form the fibrous capsule, which is an attempt to isolate the foreign material from local tissue environment (de Fougerolles and Koteliansky, 2002; Diegelmann and Evans, 2004; Ratner and Bryant, 2004). The degree and extent of the macrophages (microglia in CNS) activation, leading to FBR at the implant site, is dependent on surface properties of the biomaterial (porosity, roughness, stiffness, and chemistry), the shape of the implant, the relationship between the surface area of the biomaterial and the volume of the implant. Thus, FBR may be likely controlled by acting on those factors. The efforts put on biomaterial studies to develop a more biocompatible device have led to understand some important FBR-modulating characteristics applied to build the more recent neural interfaces: (i) the acute response is proportional to the diameter of the interface, so smaller implant size elicits less fibroblasts (astrocytes) reactivity (Szarowski et al., 2003; Veiseh et al., 2015); (ii) the adhesion of serum components and macrophages (microglia) is decreased on hydrophilic materials (Leung et al., 2008); (iii) shape without sharp corners allows for a mild mechanical trauma insertion (Groothuis et al., 2014; Veiseh et al., 2015); (iv) material requires a degree of stiffness able to give the interface the right strength to bear the tissue tension and, at the same time, the flexibility to reduce the mismatch with the surrounding tissue (Schoen and Anderson, 2004; Moshayedi et al., 2014); (v) electrical contact symmetrically distributed along the interface contribute to a uniform current distribution (Harnack et al., 2004; Groothuis et al., 2014); (vi) roughness effects are still very unpredictable since they are cell-specific and dependent on the material and porosity (Christo et al., 2015).

There are many ways by which it is possible to interfere with the FBR, but none of them seemed to solve the issue. Thus, we think it is mandatory to elucidate molecular and cellular pathways involved in the host immune response triggered by the electrode, and how drugs, co-factors and biomaterials can affect them. However, this kind of studies is very challenging due to the lack of a good, reproducible in vitro model that recapitulates what physiologically happens in the nerve tissue architecture. The 3D in vitro model developed so far to quantify inflammatory response to biomaterial might be a starting point to reproduce an immune system cells environment within the nervous system (Almeida et al., 2014; Parks et al., 2014). Other challenges are present also in the in vivo model, meaning inter-species variabilities and the difficulties of monitoring the roles of each cellular component and cytokine network involved in FBR. Moreover, all the issues related to how to specific deliver, locally or systemically, any kind of bio-active anti-inflammatory substance over the time should be analyzed. This review highlights the development of strategies with regulating effect on the FBR, with important implications for PNS stimulation and for the possibility to create implanted long-term functional medical devices. We propose a systematic approach for FBR by linking each step of the FBR process (i.e., blood and plasma proteins adsorption to the foreign body; monocyte recruitment and differentiation to macrophages and their fusion to form giant cells; fibroblasts recruitment and activation to form fibrotic tissue) with possible solutions. These solutions, of which pros and cons are underlined, are a selection, from literature and the related research field, of what we think is the most promising strategy. Based on these premises, we focus on and provide two main directions to overcome FBR and promote interface stability: (1) acting on the interface properties and (2) acting on the interface-microenvironment interaction.

Over the last years, the use of intraneural interfaces has increased in a wide range of applications, such as the control of a new generation of neurally-interfaced prostheses. At present, the success of this technology is limited by an inefficient transduction of the electrical signal, mostly due to the formation of a fibrotic tissue around the implant caused by the FBR. There are many ways by which it is possible to interfere with the FBR, but none of them seems to have solved the issue so far. Starting from recent developments in biomaterials and from novel approaches to inflammatory/fibrotic pathologies, in this review, we have explored and selected the biological solutions that might be adopted in the neural interfaces FBR context.

In particular, we have focused on and provided two main directions to overcome FBR and promote interface stability: acting on the interface properties and acting on the interface-microenvironment interaction. Hitherto, we tried to objectively highlight pros and cons of each approach and, in the following line, we critically state which one, in our opinion, seems the most promising.

Acting on the interface properties, meaning modifying porosity, roughness, stiffness, chemistry with specific surface coatings, is probably the most efficient and feasible strategy, since it modulates the FBR from the initial plasma protein absorption to the interface, the first trigger of the FBR cascade. These coatings are long-term chemically stable, small enough to not alter the surface shape, biocompatible; they can be very easily functionalized and their chemical and electrical properties can be improved. Having the competence and the suitable instrumentation, it is more feasible to modify a material physically and chemically and then directly test it in vivo, than having to deal with the choice of a specific target of the complex downstream activated pathway. Indeed, acting on the interface-microenvironment foresees the use of drugs that target the complicated immune system components or the fibroblast cell population. This option, even if doable, requires a more comprehensive evaluation and a more complex set up: i) a good, reproducible in vitro model that recapitulates what physiologically happens in vivo-so far, there are few 3D in vitro model, still not consistent enough, but they might be a starting point to reproduce an immune system cell environment within the nervous system-; ii) the choice of dose and how to specifically deliver any kind of anti-inflammatory/fibrotic substance over the time in order to avoid cytotoxic effects. The choice of a systemic delivery has undesirable long-term and wide-range effects, while topic delivery from specific scaffolds has a limited lifetime. Local release with a tested concentration appears to be a better solution, since it elicits a more specific and less toxic response in the area surrounding the interface. The issue of refilling the scaffold can be overcome by specific technologies for a long-term release, even if it adds more variables to the system. In our opinion, acting on the interface-microenvironment is more difficult to handle and evaluate, is less stable over time, produces more undesirable effects and has less impact on the FBR.

However the possibility of developing a futuristic material to promote electrode bio-integration could open new ways toward the immune response modulation strategy. A promising approach may come from the development of “bio-scaffolds” (Struzyna et al., 2015).

The novelty of this type of scaffolds is the incorporation of living cells (Eberli and Atala, 2006; Korecka et al., 2007), with the main purpose of the creation of a biological interface covering and hiding the electrode, so that the immune system does not recognize it as non-self body (Cullen et al., 2011a,b; Cullen and Smith, 2013) and does not trigger the inflammatory response. The scaffold, placed on an intraneural electrode, can provide a 3D, anisotropic biomaterial structure for different types of cells. A futuristic approach may even see the use of stem cells or autologous nervous cells to avoid host tissue inflammation and over-engineered cells that secrete factors to completely inhibit some FBR driven pathways and promote physiological integration, overcoming the lifetime limitation of soluble factors loaded or coated scaffolds (Eberli and Atala, 2006; Madduri et al., 2010; Rosenstein et al., 2010). Moreover, in case of nerve fiber damage, demyelination and axon retraction, which is a risk that we cannot exclude during the insertion of the intraneural electrode, these scaffolds could secrete trophic factors (Korecka et al., 2007), such as nerve growth factor, glial derived neurotrophic factor, insulin growth factor (Fine et al., 2002; Apel et al., 2010; Madduri et al., 2010; Yan et al., 2012), that mimic the microenvironment, helping cells to restore the damaged nerves (Thompson et al., 2016). Hitherto, in order to create these bio-scaffolds, there are few novel technologies available: (i) 3D printing, that has the advantage of having control over stiffness, pore size and integration of soluble factors and cells more closely resembling the in vivo environment (Lee et al., 2014); (ii) “cells electrospinning,” for generation of cell fibers integrated in a scaffold (Arumuganathar and Jayasinghe, 2008); (iii) intraneural electrode with cells electrospinned microchannels coupled with an already existing micro-fluidic technology (Takehara et al., 2014), developed for low invasive biocompatible installation and drug delivery.

The analysis done so far is comprehensive, but the insights we can gather on the successful strategy to overcome intra-peripheral nerve electrode FBR are limited by the following issues: (a) most of the studies on the topic were performed in the CNS; (b) most of the long-term experiments with intraneural electrodes were performed in rats that exhibit different inflammatory response and fibrotic reaction compared to human, also due to the different electrode/nerve size ratio; (c) only one experiment, which lasted <30 days, was done in a large animal (pig); (d) the few implants of intraneural electrodes in human subjects published so far did not established a direct correlation between functional data and FBR. In order to be able to manage intraneural electrodes and deeply understand their real efficacy and clinical applicability, further long-term implants are warranted.

Despite the likely impact of FBR on the long-term implant functionality, it is indeed not possible to estimate the exact incidence or provide statistical data. This is due to the lack of a consistent number of systematical studies involving a coherent group of animals, significant time-points and a reliable general method to measure FBR.

For our scope, how to achieve a reliable measure of FBR is a pertinent topic, but actually, there are no standardized scales; here we summarize the approaches used so far in vitro and in vivo. In vitro, it is possible a direct quantification on fixed tissues from animal model, where macroscopic and microscopic analyses can be made. The general morphological evaluation of the nerve and the capsule surrounding the implant is performed through the histological analysis of the nerve sections, while the grade of inflammatory response is evaluated by immunohistochemistry (IHC) for macrophages and the impact of fibrosis on nerve fiber density by IHC for axons and myelin markers.

Today, only indirect FBR measures have been used in vivo, such as the impedance of the electrode contacts which, despite being influenced by multiple factors, correlates with fibrotic tissue development. Indeed, a repeated monitoring of impedance spectrum change can give interesting information on FBR development over time.

Moreover, linked with an impedance increase, FBR alters nerve conduction. Nerve conduction studies provide electrophysiological parameters (amplitude and latency of motor and sensory evoked responses: compound motor action potential-CMAP- and compound nerve action potential-CNAP) that are considered good predictors of an ongoing FBR (Badia et al., 2011; Wurth et al., 2017). In an in vivo human application perspective, it would be also very useful to improve some existing non-invasive imaging techniques to gather structural data on FBR, such as echography, with a spatial resolution good enough to discriminate and evaluate the small area around the nerve implant.

A final important consideration is that, compared to other implant devices, neural interfaces have the peculiarity to deliver currents. It is still matter of debate whether this is relevant for FBR development. So far, the only data correlating electrical stimulation with a side effect on surrounding tissues came from brain studies where intracortical stimulating electrodes caused a thicker layer of fibrosis, compared to the non-stimulating ones (Dauth et al., 1977). The size of the lesion, the inflammatory response and the final fibrosis influence local electrical field, charge density and duration of stimulation, due to electrochemical reactions and overheating at the implant site (van Kuyck et al., 2007). However, in a recent publication it has been reported that electrical stimulation does not primarily contribute to enhance the FBR, thus insertion trauma and chemical/mechanical mismatch represent the major players in this process (Wurth et al., 2017). Therefore, even if all the intraneural technology uses a safe range of electrical parameters, thus allowing a prospective clinically-safe stimulation and recording process, it is very important to deeply understand the processes affecting the FBR-stimulation relation. To present knowledge, despite we cannot exclude a peculiar effect of current delivering on post-implantation tissue response, there is not enough scientific evidence to exploit the modulation of the stimulation parameters (waveform and frequency) as an additional strategy to reduce FBR.

FL designed the paper, analyzed the literature and wrote the paper. FR and GD designed the paper, supervised the writing, wrote and revised the paper. GV, LZ deeply revised the paper. All the authors read and approved the manuscript.