Definition: surgery may, either systematically or otherwise, cause an injury to a clearly identified nerve. In some cases, direct surgical exploration has made it possible to identify the nerve injuries.

The best studied operation is definitely thoracotomy: separation of the ribs under tension is accompanied as a minimum by crushing of the intercostal nerve which is directly involved, but through a pressure transmission effect, also of the adjoining nerves both superior and inferior to it; this has been objectively shown by a perioperative conduction block [1].

Operations involving a low transverse laparotomy (a Pfannenstiel incision or similar), as used for hysterectomy or Caesarean section, may involve the iliohypogastric and ilioinguinal nerves, which supply the skin below the lateral margins of the incision; rather than nerve sections, in this case entrapment within the reactive fibrosis is most commonly suspected [2, 3].

Inguinal herniorrhaphy, an operation carried out within a territory which is smaller but with denser innervation and more anatomically complex, exposes the iliohypogastric and ilioinguinal nerves to the risk of injury, along with the genital branch of the genitofemoral nerve, as well as the lateral cutaneous femoral nerve [37–45]; again nerve sections are implicated and may result in neuromas or entrapments [43, 46]. This phenomenon has probably been amplified by changes in practice: 15 years ago, the direct approach without mesh implantation (the Shouldice procedure) was still in common use [47] but the direct approach using a tension-free restraining mesh has now become much more widespread at the expense of the earlier technique, which is considered to be more frequently complicated by recurrent hernia [48, 49]. These restraining meshes have the effect of causing a reactive fibrosis.

In breast surgery, it is the axillary dissection that exposes small nerve endings to injury—and these are indeed sometimes difficult to identify—particularly because this dissection involves a wide area (radical clearance). These nerves are the intercosto-brachial nerve [50], the long thoracic nerve and the thoracodorsal nerve [51]. Cutaneous branches of the intercostal nerves may be involved in true mastectomy.

In knee surgery, a number of different types of nerves are exposed to injury or damage:

During sternotomy, the anterior nerve endings of the intercostal nerves are theoretically exposed; the risk seems to be greater when the internal thoracic artery is dissected away from the deep surface of the sternum [70, 71].

Clear anatomical evidence exists in the case of saphenectomy (saphenous nerve) [72], mandibular osteotomy (lingual and inferior alveolar nerves) [73–75] and anterior sampling of iliac bone grafts (lateral cutaneous nerve of the thigh), a procedure carried out in the context of reconstructive operations, mostly on the spine or mandible [76]. These factors are more speculative in the case of other abdominal operations; the nerve endings at the lateral margin of the right latissimus dorsi muscle may be injured by the initial incisions made during a laparoscopic cholecystectomy [77, 78]. More generally, a midline incision such as the one used for the majority of laparotomies is only at risk of injuring the terminal branches, which have a very small afferent cutaneous territory.

Definition: the signs observed in patients suffering from CPSP often suggest a neuropathic process. The symptoms may be purely subjective (spontaneous burning or vice–like pains, very brief paroxysmal pains like electric shocks or stabbing, a sensation of having a “missing part” or phantom part or even numbness), and are ascertained through open or focused questioning; other signs are observed or elicited through clinical examination (hypoaesthesia, dynamic allodynia, Tinel’s sign etc.) [32, 79]. Signs suggestive of a peripheral neurological abnormality remote from the surgical territory without details of the pain are a more indirect form of evidence.

The operations carried out for breast cancer are the best documented of all in this area [14, 15, 50, 80–92]. Three syndromes have been identified [13]: intercosto-brachial neuropathy, neuroma within the scar, and the sensation of having a phantom breast after total mastectomy [16, 93–95]. Non-cancer surgery—although it carries a lower risk of nerve injury—is not without risk of CPSP [96].

The case of thoracotomy is also well documented [97–106], but the non-neuropathic component no doubt still has an important place in the aetiology of some cases of CPSP [12, 105].

Inguinal herniorrhaphy is the third operation which is clearly suspect on the grounds of signs and symptoms [45, 87, 107–118].

Sternotomy may cause suggestive symptoms, with suspicions generally being raised on the basis of global but detailed tools such as the McGill Pain Questionnaire, and/or the discovery of areas of presternal hypoaesthesia [87, 119–125]. Precise diagnostic tests, specifically including anaesthetic blocks, have been carried out in patients with chronic pain [70, 71].

Clinical signs and symptoms have also been found after:

After knee arthroscopy, the clinical features seen often include sensory deficits, whether the operation was anterior cruciate ligament reconstruction using patellofemoral tendon graft [25, 58, 62] or sampling from the pes anserinus [24, 26, 27, 135], or meniscectomy [29, 30, 53, 67]. It is possible, however, that pain on kneeling is caused by a dynamic proprioceptive allodynia in the sub-patellar region [27].

Definition: functional investigations carried out within the surgical territory and/or painful area in cases of CPSP have shown changes suggestive of a mononeuropathy.

The oldest functional investigations (electromyography [EMG], somatosensory evoked potentials [SEP]), have been carried out after thoracotomy and have demonstrated persistent dysfunction of the intercostal nerve and the link between this and CPSP [136, 137]. Since then, these techniques—which are difficult to carry out mainly because the signal is difficult to reproduce—have not been used very often in studies of CPSP; changes in inferior alveolar nerve conduction have, however, been found after mandibular osteotomy [73, 74]. A real association between functional changes and CPSP is still difficult to demonstrate, due to the small numbers of patients included in the studies.

The development of quantitative methods for sensory abnormalities (Quantitative Sensory Testing or QST) makes it possible to provide evidence of sensory alterations based on specific types of stimulus (mechanical, vibration, thermal etc.) and to correlate these with the clinical data for larger numbers of patients [138]. These investigations have been carried out after certain operations that cause CPSP (inguinal herniorrhaphy, mastectomy, mandibular surgery and thoracotomy) and have shown changes in the function of peripheral fibres and/or the perception of certain sensations when stimulating the area around the scar [74, 103, 106, 139–142]. Quite surprisingly, a “standard profile” appears to emerge, with lowered mechanical thresholds and raised heat thresholds. Two comments can be made at this stage:

Only microneurographical studies, however, will be able to provide support for these hypotheses.

It should be noted that these functional changes are not necessarily linked to the occurrence of pain [145], although an association is often found [74, 103, 106, 141]. It may, of course, be suggested that psychological aspects play a part in influencing the expression of pain, but not in isolation [146, 147].

To illustrate this point, in a study of 69 patients who were examined 4 months after thoracotomy, we correlated various signs from both clinical examination and QST carried out within the surgical territory; all the criteria were bimodal nominal variables (present/absent), and numerical variables had been transformed beforehand depending on their position relative to the median. Multiple Correspondence Analysis (MCA) was carried out, and the modalities were then grouped into an ascending hierarchical classification depending on their MCA coordinates [103, 106]. The quality of the model was good overall, and 79.2 % of the global variation was represented on the first two axes.

Three groups of modalities (clusters) were identified:

This confirms the idea of a mononeuropathy which does not necessarily cause pain, illustrated by the “raised thermal threshold” group (no. 2), which is therefore distinct from the group that contains painful components (no. 1) and the group that could be classified as non-pathological (no. 1).

Definition: a preventative or curative intervention which is considered to be specifically effective for a mononeuropathy, has been proven to be effective in certain types of CPSP.

Without minimising the role of other mechanisms that have been implicated as a cause of CPSP, the purpose of this review is to highlight the importance of surgical nerve injury. It also appears that research should focus on the initial nerve injury from both anatomical and mechanistic perspectives. This knowledge will make progress possible towards validated techniques that can be used for surgical prevention. Drug treatments—either preventative or curative—could be given and could yield better results if better preclinical models that mimic postsurgical mononeuropathy are developed beforehand.