Вы находитесь на странице: 1из 10

Clinical States/

Cancer Pain




Cancer Pain: Causes,

Consequences, and Therapeutic
Patrick W. Mantyh

Similar to cancer itself, the factors that drive cancer
pain evolve and change with progression of the disease. Surgery, radiation therapy, and chemotherapy
are used to remove or kill cancer cells, and all can
induce pain and/or dysfunction of sensory and sympathetic nerve fibers. In cases in which the cancer
continues to grow or relapse occurs, cancer cells and
their associated stromal cells generate ongoing pain
by releasing algogenic substances, including protons,
bradykinin, endothelins, prostaglandins, proteases,
and tyrosine kinase activators. With disease progression, tumor growth can directly injure nerve fibers
and thereby give rise to neuropathic pain. Additionally, recent studies have demonstrated that cancer
and its associated stromal cells release tyrosine kinase
activators, including nerve growth factor, which can
induce active and highly pathological sprouting and
neuroma formation by sensory and sympathetic nerve
fibers. This structural reorganization of sensory and
sympathetic nerve fibers along with cellular and neurochemical reorganization in the spinal cord and brain
may contribute to breakthrough pain. Incorporating
this new understanding of cancer pain into novel therapies to treat cancer pain may increase the survival,
quality of life, and functional status of cancer patients
and survivors.


In 2010 cancer will be diagnosed in more than 12 million
people worldwide (excluding non-melanoma skin cancer),

and 8 million individuals will die of cancer (Boyle and Levin

2008). Cancer incidence rates are stable or slightly falling
in developed countries, whereas in developing countries,
these rates are increasing because smoking, obesity, and
increased life expectancy have led to a rapid rise in the incidence of cancer (Boyle and Levin 2008, Khan et al 2010,
Jemal etal 2011). Additionally, as detection and treatment
of most cancers have dramatically improved, survival rates
have increased so that even patients with metastatic cancer
are surviving years to decades beyond their initial diagnosis
(Jemal etal 2010).
Cancer-associated pain is one of the most common initial
symptoms that results in the diagnosis of cancer. Although
pain can be present at any time during the course of the disease, it generally increases with disease progression such that
7590% of patients with metastatic or advanced-stage cancer
will experience significant cancer pain (van den Beuken-van
Everdingen etal 2007, Costantini etal 2009). Figure 72-1 outlines some of the mechanisms that drive cancer pain, including
cancer therapy, factors released from tumors that sensitize or
excite primary afferent neurons, tumor-induced nerve injury,
and tumor-induced nerve sprouting and neuroma formation (Mantyh 2006, Gordon-Williams and Dickenson 2007,
Jimenez-Andrade etal 2010a, Mantyh etal 2010).
In patients with cancer, undergoing surgery or receiving
the full dose of radiation therapy or chemotherapy is one of
the most significant factors in determining patient survival
(Mantyh 2006). However, peripheral nerve neurotoxicity and
the accompanying pain are major side effects of radiation
therapy and many of the most commonly used antineoplastic agents, including the taxanes (e.g., paclitaxel, docetaxel),
vinca alkaloids (e.g., vincristine, vinblastine), platinum-based
compounds (e.g., cisplatin and oxaliplatin), and proteasome
inhibitors (e.g., bortezomib, disulfiram) (Quasthoff and
Hartung 2002, Cata etal 2006, Mantyh 2006, Mielke etal
2006, Bennett 2010). If the chemotherapy-induced peripheral

- Release of algogenic
factors by tumor/stromal
Cancer treatmentrelated pain


- Sensitization of nociceptors
by growth factors or cytokines
- Tumor-induced nerve injury

Pain (Intensity)

- Ectopic nerve sprouting

induced by growth factors
- Central sensitization


1030 Section Eight | Clinical States/Cancer Pain

Time (Disease Progression)

Figure 72-1. An evolving set of mechanisms drives cancer pain. Cancer pain may initially be directly related to treatment of the cancer (i.e., diagnostic

or therapeutic surgical procedures such as biopsy, resection) or be a side effect or toxicity related to the therapies used to treat cancer. Early ongoing pain is
driven primarily by factors released from cancer cells and their associated stromal cells. Cancer pain becomes more intense as nociceptors are injured by the
growing tumor and become sensitized by release of factors such as nerve growth factor, which can be released by cancer and stromal cells. With progression
of cancer, the severity of the pain tends to increase as spontaneous and movement-evoked pain/breakthrough pain occurs. This breakthrough pain may in part
be due to tumor- and stromal-induced pathological sprouting of nociceptors and the formation of neuroma-like structures. Breakthrough pain also appears
to be due in part to changes in the central nervous system, which include neurochemical and cellular reorganization of the spinal cord, as well as enhanced
synaptic transmission mediated through A and C fibers in the substantia gelatinosa of the spinal cord.

neuropathy becomes severe enough, the oncologist or patient

may reduce or cease chemotherapy treatment, which decreases
the survival rate of the patient and the likelihood that the
patient will be disease free.
In cases in which the tumor is inoperable or relapse occurs,
cancer and its associated stromal cells can induce significant
pain. This pain can arise from the original site of the cancer
(i.e., pancreatic cancer, head and neck cancer, osteosarcoma)
(Dreghorn etal 1990, Zhu etal 1999, Lam and Schmidt 2011)
or from distant sites (such as bone, liver, and lung), where
common cancers such as breast, prostate, kidney, and lung
cancer avidly metastasize (Coleman 2006). The original quality of the tumor-induced pain is usually described as dull in
character, constant, and gradually increasing in intensity with
time (Dy etal 2008) (see Fig. 72-1). If the disease continues,
a second type of cancer pain known as breakthrough or
severe incident pain can emerge (Mercadante 1997). Incident or breakthrough pain, which is defined as a transitory
flare of extreme pain superimposed on an otherwise stable
pain pattern in patients treated with opioids (Casuccio etal
2009), can occur spontaneously or with movement or weightbearing of a tumor-bearing organ or tissue (Mercadante etal
2004). Since breakthrough pain is frequently acute and unpredictable in onset, this pain can be severe, debilitating, and difficult to fully control (Coleman 1997, Mercadante 1997).
Currently, tumor-induced pain is largely managed with an
analgesic ladder that was originally promulgated by the
World Health Organization in 1986 (see Chapter 75). This
ladder begins with a non-steroidal anti-inflammatory drug
(NSAID); if the pain worsens, an NSAID plus a mild opiate;
and finally, when the pain becomes severe, an NSAID plus a
strong opiate. In addition to this three-step ladder, other adjuvant therapies, including radiation therapy, radioisotopes,
nerve blocks, nerve lesions, antiepileptics (e.g., gabapentin, carbamazepine), antidepressants (amitriptyline, imipramine), and steroids, are commonly used to control cancer

pain (Desandre and Quest 2009). It should be stressed that

most cancer pain can be controlled if it is closely monitored
and these therapies are used in a timely and proactive manner. However, the abovementioned therapies all have significant unwanted side effects (Montagnini and Zaleon 2009),
and closely monitoring and fully controlling the cancer pain
(especially if breakthrough pain is present) can be very timeconsuming for the patient, caregiver, and physician (Lossignol
and Dumitrescu 2010). Developing new analgesic therapies
that are efficacious and have fewer side effects than current
analgesics do and incorporating these advances into mainstream cancer therapy will significantly improve the quality of
life and functional status of both the patient and the caregiver.


Given the enormous consequences in terms of human suffering that cancer pain can cause, it was surprising to us when
we first began exploring the mechanisms driving cancer pain
that no well-established animal model was available for
studying any form of cancer pain. However, there were two
commonly used invivo models to study tumor-induced bone
destruction. In the first model, tumor cells are injected into
the left ventricle of the heart and then spread to multiple sites,
including the bone marrow, where they multiply, grow, and
destroy the surrounding bone (Arguello et al 1988, Yoneda
et al 1994). Although this model replicates the observation
that most tumor cells metastasize to multiple sites, including
bone, a major problem with the model is animal-to-animal
variability in the sites, size, and extent of the metastasis. Since
the tumors frequently metastasize to vital organs such as the
lung or liver, the general health of the animal is also variable, which makes behavioral assessment difficult. Additionally, because the tumors frequently metastasize to bone in the
vertebral column, tumor growth in the vertebrae can result

Chapter 72 | Cancer Pain: Causes, Consequences, and Therapeutic Opportunities 1031


Figure 72-2. Progressive destruction of mineralized

bone in mice with bone cancer. A, Low-power anteroposterior radiograph of the mouse pelvis and hindlimbs after
unilateral injection of sarcoma cells into the distal part of
the femur and closure of the injection site with an amalgam
plug (arrow), which prevents the tumor cells from growing outside the bone; arrowheads indicate areas of significant tumor-induced bone destruction of mineralized bone.
B, Radiographs of murine femora show the progressive
loss of mineralized bone caused by tumor growth. These
images are representative of the stages of bone destruction
in the murine femur. At week 1, there is minor loss of bone
near the distal head (arrowhead); at week 2, substantial
loss of mineralized bone at both the proximal and distal
(arrowhead) heads; and at week 3, loss of mineralized bone
throughout the entire femur and fracture of the distal head
(arrowhead). Scale bar = 2 mm. (Modified from Schwei MJ,
Honore P, Rogers SD, etal 1999 Neurochemical and cellular reorganization of the spinal cord in a murine model
of bone cancer pain. Journal of Neuroscience 19:10886
10897. Copyright 1999 by the Society for Neurosciences).

in collapse of the vertebral column and compression of the

spinal cord with resultant spinal dysfunction and paralysis.
Given these problems, development of a model of bone cancer
pain using intracardiac injection proved difficult at best.
The second major model used to study tumor-induced
bone destruction involved the direct injection of lytic sarcoma cells into the intramedullary space of the mouse tibia
or femur. The major problem with this model was that since
the injection site could not be plugged with conventional
sealing agents (because it is a wet, bony surface), the tumor
cells rapidly escape and grow avidly in nearby skin and
joints, thereby resulting in large extraskeletal tumor masses
that not only interfered with behavioral analysis but also
destroyed nerves passing though these sites and generated a
neuropathic pain state. We chose to adapt and modify this
model by plugging the injection hole with a dental amalgam
that tightly binds and seals the injection hole in the proximal
head of the femur. Plugging of the injection site allowed us
to contain the tumor cells within the intramedullary space
and prevented invasion of tumor cells into the surrounding
soft tissue (Fig. 72-2) (Honore et al 2000). This advance,
along with techniques with which we could simultaneously
measure bone cancerinduced pain behavior, tumor growth,
and tumor-induced bone remodeling has provided us with

the first preclinical cancer pain model, which we then used

to define the mechanisms that generate and maintain bone
cancer pain.
After injection and confinement of primarily osteolytic
2472 murine osteosarcoma tumor cells to the intramedullary
space of the mouse femur, the tumor cells grow in a highly
reproducible fashion as they proliferate and replace the hematopoietic cells in the bone marrow (Schwei etal 1999, Sabino
etal 2002). Eventually, the entire marrow space is homogeneously filled with tumor cells and tumor-associated inflammatory/immune cells. In terms of bone remodeling, injection
of osteosarcoma cells into the femur induces a dramatic
proliferation and hypertrophy of osteoclasts at the tumor
bone interface, with significant bone destruction in both the
proximal and distal heads of the femur (see Fig. 72-2). In the
osteosarcoma model, ongoing pain and movement-evoked
pain-related behavior increased in severity with time, and
this pain-related behavior correlated with the tumor growth
and progressive tumor-induced bone destruction, which mirrors what occurs in patients with primary or metastatic bone
cancer. Although sarcoma cells constituted the tumor used in
the first bone cancer pain model that we developed, we have
since developed other bone cancer pain models using prostate,
breast, melanoma, colon, and lung tumors, all of which have

1032 Section Eight | Clinical States/Cancer Pain

provided insight into the similarities and differences by which
different tumors drive bone cancer pain (Sabino etal 2003).

(Mantyh 2006, Joyce and Pollard 2009), many of which have

been shown to sensitize or directly excite primary afferent
neurons (Julius and Basbaum 2001).



Tumor-Induced Acidosis

A tumor is composed of not only cancer cells but also tumorassociated stromal cells. In most tumors, stromal cells far
outnumber cancer cells and include endothelial cells, fibroblasts, and a host of inflammatory and immune cells, including macrophages, mast cells, neutrophils, and T lymphocytes
(Joyce and Pollard 2009) (Fig. 72-3). Both cancer cells and
their associated stromal cells secrete a wide variety of factors

The finding that a subpopulation of sensory neurons express

transient receptor potential vanilloid 1 (TRPV1, which is also
known as the capsaicin receptor) and the acid-sensing ion
channel 3 (ASIC3) and that both these channels respond to
acidosis is of significant interest to researchers studying cancer
pain (Joyce and Pollard 2009) because cancer cells in general
have a lower pH (6.8) than normal cells do (pH 7.2) (Griffiths

Unmyelinated fibers (C)
and thinly myelinated (A)

tissue (bone)

Spinal cord

Tumor/stromal cells

Tumor-associated immune cells

Macrophage Mast cell

T cell


Growth factors




B2 R






Nav 1.7
1.8, and 1.9


H+ H+



Figure 72-3. Primary afferent sensory nerve fibers and the generation and maintenance of cancer pain. Primary afferent neurons innervating the
body have their cell bodies in the dorsal root ganglia (DRG) and transmit sensory information from the periphery to the spinal cord and brain. Unmyelinated
C fibers and thinly myelinated A fibers contain small-diameter cell bodies that project centrally to the superficial spinal cord. These fibers are involved in
detecting multiple noxious stimuli (chemical, thermal, and mechanical). Box: Nociceptors use several different types of receptor to detect and transmit signals
about noxious stimuli produced by cancer and stromal cells (yellow), tumor-associated immune cells (blue), or other aspects of the tumor microenvironment.
Multiple factors may contribute to the pain associated with cancer. Transient receptor potential vanilloid 1 (TRPV1) and acid-sensing ion channels (ASICs)
detect the extracellular protons produced by tumor-induced tissue damage or abnormal osteoclast-mediated bone resorption. Several mechanosensitive ion
channels may be involved in detecting the high-threshold mechanical stimuli that occur when distal aspects of the sensory nerve fiber are distended by mechanical pressure as a result of the growing tumor or as a result of destabilization or fracture of bone. Tumor cells and associated inflammatory (immune) cells
produce a variety of chemical mediators, including prostaglandins (PGE2), nerve growth factor (NGF), endothelins (ET), bradykinin, (BK) ,and extracellular
adenosine triphosphate. Several of these pro-inflammatory mediators have receptors on peripheral terminals and can directly activate or sensitize nociceptors.
NGF, together with its cognate receptor TrkA, may serve as a master regulator of bone cancer pain by modulating the sensitivity and increasing the expression
of several receptors and ion channels that contribute to the increased excitability of nociceptors in the vicinity of the tumor.

Chapter 72 | Cancer Pain: Causes, Consequences, and Therapeutic Opportunities 1033

1991). Importantly, many tumors that metastasize to bone
induce a marked proliferation and hypertrophy of osteoclasts. Osteoclasts avidly resorb bone by generating a pH of
24 in their resorption bay, which drives the excessive bone
resorption that can ultimately lead to fracture of the tumorbearing bone (Clohisy et al 2000). To test whether TRPV1
channels were expressed by sensory nerve fibers that innervate
tumor-bearing tissue and whether TRPV1 contributed to cancer pain, an invivo model of bone cancer pain was explored
(Ghilardi etal 2005). In these studies it was shown that a subpopulation of nerve fibers that innervate the tumor-bearing
bone express TRPV1, that acute or chronic administration of
a TRPV1 antagonist attenuates bone cancer pain, and that
disruption of the TRPV1 gene results in attenuation of bone
cancer pain (Ghilardi et al 2005). Furthermore, administration of a TRPV1 antagonist to TRPV1 null animals with bone
cancer resulted in no further reduction in the pain that was
already present in the TRPV1 null mice (Ghilardi etal 2005).
To date, the results of human clinical trials with TRPV1 or
ASIC3 channel antagonists have not been reported in subjects
with cancer pain. However, as discussed later, understanding
the role that TRPV1 plays in driving bone cancer pain has
provided insight into why therapies that inhibit osteoclasts,
including bisphosphonates and denosumab (an anti-RANKL
fully humanized monoclonal antibody), are efficacious in
reducing bone cancer pain (von Moos et al 2008, Stopeck
etal 2010, Henry etal 2011).
The skeleton is the most common site for distant metastasis from prostate, breast, thyroid, lung, and renal carcinoma
(Coleman 2006). Once tumor cells have metastasized to bone,
a cycle of tumor growth, bone destruction, and the formation
of woven bone begins and results in significant pain, skeletal
fractures, and hypercalcemia (Coleman 2006). Cancer cells
themselves do not destroy bone but rather they and their associated stromal cells express the receptor activator of nuclear
factor B ligand (RANKL), which binds to the receptor RANK
expressed by osteoclasts. Activation of the RANKL/RANK
pathway promotes the proliferation and hypertrophy of these
bone-destroying osteoclasts (Clohisy and Mantyh 2004).
Osteoclasts resorb bone by forming a highly acidic resorption
bay or pit between the osteoclast and bone, which can
stimulate TRPV1 or ASIC3 channels and drive bone cancer
pain (Clohisy and Mantyh 2004). In the past decade multiple
studies have shown that two therapies that reduce osteoclast
function also significantly reduce bone cancer pain (Honore
etal 2000, Lipton 2008, von Moos etal 2008, Stopeck etal
2010, Henry etal 2011).
The first and most widely used therapy is the class of compounds known as bisphosphonates, which avidly bind to
bone. Once the bisphosphonate has bound to bone, osteoclasts that are resorbing bone generally need to actively take
up the breakdown products of bone at the apical (bone facing)
surface and transfer these products via transcytosis for release
at the distal surface of the osteoclast so that they can be disposed by exocytosis (Stenbeck 2002). However, if a bisphosphonate is tightly bound to the bone that is being resorbed,
the bisphosphonate will also be taken up by endocytosis
(Rogers et al 2000). Once internalized, the bisphosphonate
interferes either with adenosine triphosphate energy metabolism (nonnitrogen-containing bisphosphonates) or with the
mevalonate pathway (nitrogen-containing bisphosphonates),
which results first in osteoclast dysfunction and ultimately in

osteoclast apoptosis (Rogers etal 2000, Clezardin etal 2005).

Because a significant population of nerve fibers that innervate
bone express TRPV1 (Ghilardi et al 2005), one way that
bisphosphonates appear to relieve bone pain is by decreasing
osteoclast-induced acidosis, which in turn will decrease activation of the ion-sensing TRPV1 or ASIC3 receptors that are
expressed by sensory nerve fibers.
Another method that is highly effective in reducing tumorinduced osteoclast bone resorption in both animals and
humans is interference in the binding of RANKL to RANK,
which is required for osteoclast proliferation and maturation
(Lipton and Jun 2008). Within 2 days of administration of
therapies that interfere with binding of RANKL to RANK
(such as osteoprotegerin or denosumab), there is an almost
complete loss of activated osteoclasts, a marked reduction in
plasma markers of bone resorption, and significant attenuation of bone cancer pain in a mouse model of bone cancer
pain (Honore etal 2000).

Tumor-Induced Mechanical Instability

of Bone
Therapies that inhibit osteoclast-induced bone resorption
also maintain the mechanical strength of bone even though
tumor cells are present in the bone. Thus, in addition to acidosis, excessive tumor-induced osteoclast bone resorption
destroys bone and leads to mechanical instability and fracture of bone, which causes mechanical distortion of the nerve
fibers innervating the bone (Yates and Smith 1994, JimenezAndrade etal 2007). Thus, following significant weakening
or fracture secondary to tumor-induced bone remodeling,
significant movement-evoked pain can occur, presumably
because of mechanical distortion of the mechanosensitive
sensory nerve fibers that innervate the bone. Clearly, pain
associated with the fracture is attenuated if the bone is stabilized and repositioned into its normal orientation (Rubert
and Malawer 2000). Both osteolytic and osteoblastic tumors
induce loss of the mechanical strength and stability of mineralized bone (Arrington etal 2006), so with significant bone
remodeling, normally innocuous mechanical stress can now
result in distortion and activation of the mechanosensitive
nerve fibers that innervate the bone. Since bisphosphonates
and anti-RANKL therapies reduce tumor-induced osteoclast
bone remodeling, preserve the mechanical strength of bone,
reduce bone fractures, and decrease osteoclast-induced acidosis in both animals and humans, these therapies are highly
useful in managing pain resulting from metastasis of cancer
to bone.

Factors Released by Cancer/Stromal

Cells That Drive Cancer Pain
One area that has significantly contributed to our understanding of what drives cancer pain is work examining the
factors released by tumor/stromal cells that drive cancer
pain and influence disease progression. These factors include
bradykinin, cannabinoids, endothelins, interleukin-6 (IL-6),
granulocyte-macrophage colony-stimulating factor (GMSCF), nerve growth factor (NGF), proteases, and tumor
necrosis factor- (TNF-). Because there are several recent
studies in this area, we will briefly summarize some of the
most striking research on this topic.

1034 Section Eight | Clinical States/Cancer Pain

Recent studies in animals have revealed that the cannabinoid
system plays an important role in modulating cancer pain.
It has been reported that tumor-induced mechanical hyperalgesia in the hindpaw is associated with decreased levels of
anandamide (endogenous agonist of the cannabinoid CB1 and
CB2 receptors) and increased degradation of anandamide in
the hindpaw skin ipsilateral to the tumor-bearing paw (Khasabova et al 2008). Furthermore, injection into the hindpaw
of anandamide or an inhibitor of the enzyme that degrades
anandamide reduced the tumor-induced hyperalgesia. A
recent report also demonstrated that acute and sustained
administration of a cannabinoid CB2 agonist attenuates both
spontaneous and evoked pain behavior in a bone cancer pain
model (Lozano-Ondoua et al 2010). Although these studies
suggest that the cannabinoid system plays a role in driving
cancer pain, future clinical studies are warranted to evaluate
the analgesic as well as the potential central nervous system
(CNS) side effects of these drugs.
Colony-Stimulating Factors
Human studies have shown that several non-hematopoietic
tumors secrete colony-stimulating factors that act on their
receptors expressed on myeloid cells, tumor cells, and nerve
fibers. Recently, Schweizerhof and co-authors (2009) reported
that levels of granulocyte colony-stimulating factor (G-CSF)
and GM-CSF in the lysates of bone marrow from tumorbearing mice were significantly increased in comparison to
the levels in nave mice. Additionally, GM-CSF sensitized the
nerves to mechanical stimuli, potentiated the release of calcitonin generelated peptide (CGRP), and caused sprouting of
sensory nerve endings in the skin (Schweizerhof etal 2009).
The actions of these colony-stimulating factors are mediated
by activation of their receptors (G-CSFR and GM-CSFR,
expressed in peripheral nerves innervating the tumor-bearing
tissue) since the administration of neutralizing antisera against
these receptors and the sensory nervespecific knockdown of
GM-CSF receptors reduced the tumor-induced pain behavior
(Schweizerhof et al 2009). Based on these studies, colonystimulating factors may be a potential target therapeutic to be
exploited in the cancer pain field.
Endothelin antagonists are another group of pharmacological
agents that offer promise in managing cancer pain and reducing progression of the disease. Endothelins (ET-1, -2, and -3)
are a family of vasoactive peptides that are expressed at high
levels by several types of tumors, including those that arise
from the prostate (Nelson etal 1995, 1996). Clinical studies
have shown a correlation between the severity of the pain and
plasma levels of endothelins in prostate cancer patients (Nelson etal 1995). Electrophysiological studies have shown that
ET-1 may directly sensitize or excite C-fiber nociceptors innervating the tumor-bearing tissue through activation of endothelin A receptors (ETAs) (Hamamoto etal 2008), which are
expressed by a subset of small unmyelinated primary afferent
neurons (Pomonis et al 2001). Furthermore, direct application of endothelin to peripheral nerves induces the activation
of primary afferent fibers and pain-related behavior (Davar
etal 1998, Hamamoto etal 2008). These results suggest that
endothelins play a critical role in tumor-induced thermal and
mechanical hyperalgesia. Currently, several ongoing human

clinical trials are examining the effects that ETA antagonists

have on cancer pain and disease progression.
Tumors release a variety of chemical agents that sensitize
peripheral afferent neurons, including cytokines. One of the
most studied cytokines in the pain field is the pro-inflammatory
and pro-algesic cytokine TNF-. This cytokine is produced by
inflammatory/immune cells, Schwann cells, and some tumor
cells (Beutler 1999). It has been reported that TNF- levels are
significantly increased in tumor micro-perfusates and tumor
site homogenates in comparison to the levels in nave mice or
the contralateral hindlimb (Wacnik etal 2005). Furthermore,
injection of TNF- into tumor-bearing mice results in heat
and mechanical hyperalgesia, which is blocked by administering the TNF- antagonist etanercept (Constantin etal 2008).
Another cytokine that has been suggested to be involved
in driving cancer pain is the pleiotropic cytokine IL-6. Several studies have shown that levels of this cytokine are upregulated under various pathological conditions (Poole etal
1995, Smith et al 2001, Nishimoto and Kishimoto 2006,
Rose-John etal 2006) and that direct injection (intradermally,
intramuscularly, intrathecally) of IL-6 results in mechanical
and thermal hyperalgesia (Poole etal 1995, DeLeo etal 1996,
Dina et al 2008). IL-6 is produced by different inflammatory/immune cells and by some tumor cells (Nishimoto and
Kishimoto 2006). The actions of IL-6 are mediated by binding to its specific receptor, IL-6R, which exists in both transmembrane and soluble form (Rose-John and Heinrich 1994,
Rose-John etal 2006). Binding of IL-6 triggers an association
of IL-6R with the transducer glycoprotein gp130 (Taga etal
1989). Recently, it has been reported that nociceptor-specific
depletion of gp130 results in a significant reduction in the
heat hyperalgesia induced by the subcutaneous injection
of LL2 carcinoma cells into the hindpaw without affecting
tumor growth (Andratsch et al 2009). These results suggest
that blocking TNF- or IL-6/gp130 may be a viable therapeutic opportunity for the treatment of cancer pain.
Tyrosine Receptor Kinases
One important concept that has emerged over the past decade
is that in addition to NGF being able to directly activate
TrkA-expressing sensory neurons, NGF activation of TrkA
appears to play a key role in the sensitization of nociceptors (see for review see Pezet and McMahon 2006; also see
Chapter 3). Thus, in addition to inducing rapid phosphorylation and sensitization of TRPV1, retrograde transport of the
NGF/TrkA complex to the neuronal cell body of nociceptors
induces increased synthesis of the neurotransmitters substance
P and CGRP and increased expression of receptors (bradykinin receptor), channels (P2X3, TRPV1, ASIC3, and sodium
channels), transcription factors (activated transcription factor 3 [ATF3]), and structural molecules (neurofilaments and
the sodium channelanchoring molecule p11) (for review
see Pezet and McMahon 2006). Additionally, NGF appears
to modulate the trafficking and insertion of sodium channels such as Nav1.8 (Gould etal 2000) and TRPV1 (Ji etal
2002) in sensory neurons, as well as modulate the expression
profile of supporting cells in the dorsal root ganglion (DRG)
and peripheral nerve, such as non-myelinating Schwann cells
and macrophages (Heumann etal 1987a, 1987b; Obata etal

Chapter 72 | Cancer Pain: Causes, Consequences, and Therapeutic Opportunities 1035

In light of the potential role that NGF may play in driving
bone cancer pain, therapies that block NGF or TrkA have been
examined in breast, prostate, and sarcoma models of bone
cancer pain. Interestingly, even though prostate cancer cells
did not express detectable levels of mRNA coding for NGF
(Halvorson et al 2005), in all three models of bone cancer
pain, not only was administration of anti-NGF therapy (using
an antibody that sequesters extracellular NGF) highly efficacious in reducing both early- and late-stage bone cancer pain
related behavior, but this reduction in pain-related behavior
was also greater than that achieved with the acute administration of 10 or 30 mg/kg of morphine sulfate (Halvorson etal
2005, Sevcik etal 2005). If cancer cells do not have to express
NGF for anti-NGF to have an analgesic effect, what cells
might be synthesizing and releasing NGF? Previous studies
have shown that many tumor-associated stromal cells, including macrophages, T lymphocytes, mast cells, and endothelial
cells, are capable of expressing and releasing NGF (Vega etal
2003, Pezet and McMahon 2006). Thus, therapies that target NGF or its cognate receptor TrkA may be efficacious in
attenuating the pain in other cancers such as ovarian, pancreatic, and head and neck carcinoma, in which a large number
of tumor-associated stromal cells express and release NGF.
Therapies targeting ETA receptors, NGF, and TNF- are
currently in human clinical trials in patients with cancer pain
(www.clinicaltrials.gov). Whether therapies targeting these
and the other aforementioned factors will reach clinical trials and be successful in humans in relieving cancer pain will
depend largely on issues related to safety, efficacy, and effects
on tumor growth and metastasis. Choosing which type of
cancer pain to target in clinical trials is currently a hit-or-miss
proposition because there are relatively few models of cancer pain and different types of tumors can have very unique
characteristics, as shown by chemotherapeutic agents, which
can be highly effective against one cancer and completely ineffective against another. Developing a better understanding of
what common and what unique factors drive different cancer
pain would be of enormous benefit because it would greatly
aid in defining which type of cancer pain will be most likely to
respond to targeted analgesic therapies in clinical trials.


In both the sarcoma and prostate bone cancer pain models
and a model of pancreatic cancer pain, as tumor cells invade
normal tissue, the tumor appears to first come into contact,
injure, and then destroy the very distal processes of sensory
fibers (Peters etal 2005). Thus, although sensory fibers appear
to have normal morphology at the leading edge of the tumor,
with time the sensory nerve fibers begin to display a discontinuous and fragmented appearance, thus suggesting that
following initial activation by the tumor cells, the distal processes of sensory fibers are ultimately injured and destroyed
as the invading tumor cells first proliferate and then undergo
necrosis as they outgrow the neovascularization that supports
them (Peters etal 2005). This initial tumor-induced activation
and then injury to sensory nerve fibers are accompanied by
an increase in ongoing and movement-evoked pain behavior.
Interestingly, there are several changes in the DRG, including hypertrophy of the satellite cells surrounding sensory
neuron cell bodies, up-regulation of ATF3, and macrophage

infiltration of the DRG, findings that have also been described

in other models of peripheral nerve injury and in other noncancerous neuropathic pain states (Peters etal 2005). These
data, as well as the fact that a component of bone cancer pain
is attenuated by gabapentin (which is approved for the treatment of neuropathic pain), suggests that a component of cancer pain is neuropathic in origin (Peters etal 2005).


Although tumor-induced injury has been observed in both animals and humans with cancer, an intriguing but largely unexplored mechanism by which cancer pain may be generated is
by active and pathological tumor-induced sprouting and neuroma formation. Previous studies in humans and experimental
animals have shown that inappropriate sprouting and/or neuroma formation and can lead to a change in the phenotype of
sensory and sympathetic nerve fibers, including up-regulation
and inappropriate insertion of sodium channels into the distal
tips of injured sensory neurons (Devor etal 1993, England etal
1996, Black et al 2008). These newly formed sensory nerve
fibers (which sprout in response to peripheral nerve injury)
exhibit both spontaneous and movement-evoked ectopic discharges accompanied by a pain that is both severe and difficult
to manage medically (Lindqvist etal 2000, Devor 2001, Black
etal 2008).
In a mouse model of pancreatic cancer pain (Lindsay etal
2005) and in breast (Bloom et al 2011), prostate (JimenezAndrade et al 2010a), and sarcoma (Mantyh et al 2010)
models of bone cancer there was first tumor-induced nerve
injury and then subsequent sprouting and formation of
neuroma-like structures by sensory (Fig. 72-4) and sympathetic
nerve fibers. To address what might be driving this ectopic
sprouting and neuroma formation, anti-NGF therapy was
given. It was found that sustained administration of anti-NGF
therapy largely blocked the pathological sprouting of sensory
and sympathetic nerve fibers and the formation of neuromalike structures and significantly inhibited the development of
cancer pain in this model (Mantyh etal 2010). Interestingly,
injection of canine prostate cancer cells (which do not express
detectable levels of mRNA encoding NGF) into the bone of
nude mice induces sprouting of CGRP-expressing and NF200expressing sensory nerve fibers and tyrosine hydroxylase (TH)positive sympathetic nerve fibers, and nearly all of these
sprouting nerve fibers co-express TrkA (Jimenez-Andrade etal
2010a). What is in some ways impressive about these results
is the extent of the sprouting; even in bone marrow, which
normally receives very modest innervation by sensory nerve
fibers, prostate cancerassociated stromal cells can induce up
to a 1070-fold increase in the density of TrkA-positive nerve
fibers (Jimenez-Andrade etal 2010a).
Data from these experiments suggested that a significant
proportion of this ectopic sprouting and neuroma formation is driven by NGF. Although it was originally assumed
that the majority of the NGF was derived from tumor cells,
another study using a tumor cell that did not express NGF
showed highly exuberant sprouting, thus suggesting that
NGF released from tumor-associated stromal cells could
drive this ectopic reorganization of sensory and sympathetic
nerve fibers. Interestingly, in both the pancreatic (Lindsay
etal 2005) and bone (Jimenez-Andrade etal 2010a, Mantyh

1036 Section Eight | Clinical States/Cancer Pain



Sarcoma + vehicle

Figure 72-4. Cancer and its associated

1 mm



Sarcoma + vehicle

GFP-Sarcoma cells/CGRP


50 m

stromal cells can induce nerve sprouting

and neuroma formation in the tumor-
bearing organ. Confocal images of the periosteum of the mouse bone were acquired
from whole-mount preparations, tiled, and
overlaid (to scale) on a three-dimensional
microcomputed tomography rendering of
a sham femur (A) or sarcoma femur (B),
respectively, with use of the Amira software.
Note that bone injected with green fluorescent protein (GFP)cancer cells (in green, B)
exhibits significant cortical bone deterioration and pathological reorganization of calcitonin generelated peptide (CGRP) nerve
fibers (in red) when compared with the sham
bone (A). The boxed areas in A and B correspond to the confocal images in C and D,
respectively. High-power confocal images
of non-decalcified, whole-mount preparations of the femoral periosteum from sham
(C) or sarcoma mice (D) show CGRP-positive nerve fibers and GFP-positive sarcoma
cancer cells. Similar nerve sprouting and
formation of neuroma-like structures are
also observed in breast and prostate cancer
invading bone.


10% (NF200+, TrkA)

10% (NF200+, TrkA)

15% (NF200+, TrkA)

10% (NF200+, CGRP+, TrkA+)

60% (NF200+, CGRP+, TrkA+)

20% (CGRP+, TrkA+)

< 5% (CGRP+, TrkA)

20% (CGRP+, TrkA+)

< 5% (CGRP+, TrkA)

45% (IB4/MrgprD+, TrkA)

~30% TrkA+

~80% TrkA+

Figure 72-5. The effectiveness of analgesic therapies for different types of cancer may depend on the specific population of nerve fibers that innervate the
tumor-bearing organ. This schematic illustrates the percentages and types of sensory nerve fibers that innervate the skin versus bone. The skin is innervated by
thickly myelinated A fibers (neurofilament 200 positive [NF200+], TrkA), thinly myelinated A fibers (NF200+,Trk and NF200+, calcitonin generelated
peptide positive [CGRP+], TrkA+), unmyelinated peptide-rich C fibers (CGRP+, TrkA+) and unmyelinated peptide-poor C fibers (isolectin B4 positive [IB4+]),
and Mas-related G proteincoupled receptor member D (Mrgprd+, TrkA). In contrast, the bone appears to be predominantly innervated by thinly myelinated
A fibers (NF200+, TrkA and NF200+, CGRP+, TrkA+) and peptide-rich C fibers (CGRP+ and TrkA+). In skin and bone there is also a small proportion (<5%
of the total) of unmyelinated C fibers that are CGRP+ and TrkA. The percentages and types of sensory nerve fibers innervating the skin were estimated with
data from previous studies (Bennett etal 1996, Lu etal 2001, Ambalavanar etal 2005, Zylka etal 2005, Nakajima etal 2008, Sugiura etal 2008, JimenezAndrade etal 2010b). Note that although approximately 30% of the sensory nerve fibers that innervate skin are TrkA+, more than 80% of the sensory nerve
fibers that innervate bone are TrkA+. The fact that a greater percentage of the sensory nerve fibers that innervate bone versus skin are TrkA+ may in part
explain why therapies that block nerve growth factor or TrkA show greater efficacy in relieving skeletal than in relieving skin pain. TrkA+ subpopulations
of neurons are shown in bold for emphasis.

Chapter 72 | Cancer Pain: Causes, Consequences, and Therapeutic Opportunities 1037

etal 2010, Bloom etal 2011) models of cancer pain, sprouting and nerve degeneration were not mutually exclusive, but
rather over a period of weeks to months nerve fibers were first
injured, then sprouting occurred, and then the nerve fibers
were reinjured when the tumor became necrotic because of
gradual loss of the vascular supply needed to maintain tumor
viability. Since damage to even the distal ends of peripheral
nerves can induce neuropathic pain, this process of tumorinduced sprouting and destruction of these newly sprouted
sensory and sympathetic fibers has the potential to contribute
to both movement-evoked and spontaneous breakthrough
cancer pain.
These studies suggest that tyrosine kinase activators, in
this case NGF activating TrkA, can induce a remarkable
and active reorganization of sensory and sympathetic nerve
fibers that may contribute to ongoing and breakthrough pain
(Jimenez-Andrade et al 2010a, Bloom et al 2011). Clearly,
other tyrosine kinase activators that induce sensitization in
tumor tissues in a manner similar to NGF, such as artemin
(Elitt etal 2006), G-CSF (Schweizerhof etal 2009), and GMCSF (Schweizerhof et al 2009), which have been shown to
be involved in tumor-induced sensitization, may also play a
significant role in promoting tumor-induced sprouting and
neuroma formation. Whether similar spouting/neuroma formation occurs in painful cancers such as ovarian, renal, head,
and neck cancer has yet to be explored. Previous studies have
demonstrated that tyrosine kinase receptor activation can
induce sprouting that is both rapid and profuse (Diamond
etal 1992). Importantly, tumor cells are constantly proliferating, metastasizing, undergoing necrosis, and then regrowing
at new sites. Thus, even if therapies that block NGF or TrkA
are given after tumor-induced sprouting and/or neuroma formation has occurred, these therapies will block the newly
forming nerve sprouting and neuroma formation (Fig. 72-5).
These results emphasize the evolving nature of cancer pain
and suggest that the earlier and more effective the analgesic
therapy, the greater the likelihood of being able to effectively
control both early- and late-stage cancer pain.






Figure 72-6. Cancer-induced reorganization of the central nervous system. Confocal images show the increase in the astrocyte marker glial fibrillary acidic protein (GFAP) in a mouse with bone cancer pain in the right
femur. Coronal sections of the L4 spinal cord were taken 21 days following the injection of osteolytic sarcoma cells into the intramedullary space
of the femur. GFAP is bright orange in A and green in B and C, and NeuN
staining (which labels neurons) is red. A low-power image (A) shows that
up-regulation of GFAP is almost exclusively ipsilateral to the femur with the
intraosseous tumor. A higher magnification of GFAP contralateral (B) and
ipsilateral (C) to the femur with cancer shows that on the ipsilateral side there
is marked hypertrophy of astrocytes characterized by an increase in both the
size of the astrocyte cell bodies and the extent of arborization of their distal
processes. Additionally, this increase in GFAP (green) is observed without a
detectable loss of neurons because NeuN (red) labeling remains unchanged.
These images, from 60 m-thick tissue, are projected from 6 optical sections
acquired at 4-m intervals with a 20 lens (scale bar = 200 m) (A) and
from 12 optical sections acquired at 0.8-m intervals with a 60 lens (scale
bar = 30 m) (B and C). (Modified from Schwei MJ, Honore P, Rogers SD,
etal 1999 Neurochemical and cellular reorganization of the spinal cord in
a murine model of bone cancer pain. Journal of Neuroscience 19:10886
10897. Copyright 1999 by the Society for Neuroscience.)

The majority of what we know about the mechanisms that

generate cancer pain has focused on changes in primary afferent sensory nerve fibers and sympathetic fibers that innervate
the tumor-bearing organ. However, several studies have demonstrated that animals with cancer pain also exhibit significant
pathological changes in the CNS that contribute to the generation and maintenance of cancer pain (for review see GordonWilliams and Dickenson 2007). Thus, it has been reported that
in mice with bone cancer pain there are simultaneous changes
in segments of the spinal cord that receive input from nerve
fibers innervating the tumor-bearing tissue. These changes
include simultaneous alterations in dynorphin, astrocytes (Fig.
72-6), microglia, c-Fos expression, and substance P internalization (Schwei etal 1999). Other reports have demonstrated that
in bone cancer models, pain-related behavior is accompanied
by increased expression of NR2B, an N-methyl-d-aspartate
(NMDA) receptor subunit, and IL-1 released from glial cells
and thought to facilitate pain by enhancing phosphorylation
of the NR1 subunit of the NMDA receptor (Zhang etal 2008,
Gu etal 2010). These latter results suggest that chemical mediators released from glial cells may control the amplitude of

synaptic responses in animals bearing bone cancer by changing expression levels of NMDA and -amino-3-hydroxy-5methyl-4-isoxazolepropionic acid (AMPA) receptors and their
phosphorylation (Zhang etal 2008, Gu etal 2010).
The possibility that cancer pain also involves changes in the
CNS is supported by a recent study using patch-clamp recordings from spinal cord slices with an attached dorsal root. It
was shown that tumor-bearing mice exhibit unique plastic
changes in spinal excitatory synaptic transmission mediated
through A and C afferent fibers (Yanagisawa et al 2010).
Invivo population studies in a rodent model of breast cancer
induced bone cancer pain revealed that in normal animals,
the proportion of widedynamic range (WDR) to nociceptivespecific neurons in this lamina is 26% WDR to 74% nociceptive specific whereas on establishment of cancer pain, this
ratio shifts to 47% WDR to 53% nociceptive specific. This
phenotypic shift of the superficial dorsal horn population is
accompanied by WDR hyperexcitability to mechanical, thermal, and electrical stimuli in the superficial and deep dorsal

1038 Section Eight | Clinical States/Cancer Pain

horn (Urch et al 2003), which correlates with the development of behavioral signs of pain and further suggests that an
ongoing state of central sensitization occurs with bone cancer
Other data suggest that with cancer pain, central sensitization is not confined to the spinal cord, but rather sites in
the brain stem that are involved in descending inhibition
and facilitation also show clear changes, thus implying that
descending controls likewise play a role in the maintenance of
cancer-induced bone pain (Gordon-Williams and Dickenson
2007). Together, these studies suggest that cancer pain not
only sensitizes primary afferent neurons but also induces significant reorganization within the CNS. Combining preclinical cancer pain studies with brain-imaging studies in human
cancer pain patients has the potential not only to provide
better understanding of how cancer pain is generated and
maintained but also to identify the changes in processing and
perception (i.e., discriminative versus affective component) of
cancer pain that occur in specific areas of the CNS.

The mechanisms that drive cancer pain clearly evolve and
change with disease progression. Cancer cells and their associated stromal cells can generate ongoing and breakthrough

pain. This pain appears to be driven in an additive fashion,

first by tumor and stromal cells releasing factors that sensitize and activate nociceptors, then by injury to sensory nerve
fibers, and finally by releasing growth factors that drive
ectopic sprouting of nerve fibers and neuroma formation, all
of which can contribute to central sensitization. Developing
therapies that target the different mechanisms that contribute to cancer pain has the potential to reduce both ongoing
and breakthrough pain. Moving forward, it will be important
to develop novel targeted therapies to treat cancer pain and
to ensure that relief of cancer pain becomes an integral part
of mainstream cancer therapy that is available to all cancer
patients in both the developing and developed world.
The references for this chapter can be found at www