Archives of Toxicology

, Volume 86, Issue 1, pp 1–12

The safety profile of imatinib in CML and GIST: long-term considerations

Authors

    • Royal Marsden NHS Foundation Trust
  • Ian Judson
    • Royal Marsden NHS Foundation Trust
Review Article

DOI: 10.1007/s00204-011-0729-7

Cite this article as:
Thanopoulou, E. & Judson, I. Arch Toxicol (2012) 86: 1. doi:10.1007/s00204-011-0729-7

Abstract

Imatinib mesylate is considered the standard first-line systemic treatment for patients with chronic myeloid leukaemia (CML) and gastrointestinal stromal tumour (GIST) by targeting BCR-ABL and c-KIT tyrosine kinases, respectively. Indeed, imatinib has substantially changed the clinical management and improved the prognosis of both diseases. Treatment with imatinib is generally well tolerated, and the risk for severe adverse effects is low, generally occurring during the early phase of treatment and correlating with imatinib dose, phase of disease and patient’s characteristics. This article summarises recent data on safety profile of imatinib for the treatment of CML and GIST, including long-term side effects. Prolonged treatment with imatinib in both diseases demonstrates excellent tolerability. There are few significant concerns and those that have emerged, like cardiotoxicity, have far turned out to be exaggerated.

Keywords

ImatinibCMLGISTSafetyToxicity

Introduction

Imatinib mesylate (formerly STI-571) is a 2-phenylaminopyrimidine compound designed to specifically interact with the adenosine triphosphate (ATP) binding site of several protein tyrosine kinases that became the prototype of the target treatment against both haematological and solid tumours. Indeed, imatinib dramatically changed the treatment of two completely different diseases, i.e. chronic myeloid leukaemia (CML), a haematopoietic stem cell disease (Faderl et al. 1999), and gastrointestinal stromal tumours (GIST), a mesenchymal tumour of the gastrointestinal tract, possibly arising from interstitial cells of Cajal (Miettinen et al. 1999), which, however, both exhibit similar mechanisms of disease development.

In CML, the presence of the Philadelphia chromosome (Ph), a consequence of an aberrant reciprocal translocation between chromosomes 9 and 22, results in the fusion gene BCR-ABL, which encodes a protein (p210) with deregulated leukaemogenic tyrosine kinase activity (Daley et al. 1990). The driving mechanism in GIST development is most commonly gain-of-function mutations of KIT, which encodes a receptor tyrosine kinase, that lead to aberrant kinase activity (Hirota et al. 1998). Thus, imatinib as a highly selective inhibitor of the chimeric BCR-ABL (Buchdunger et al. 1996) and KIT proteins (Joensuu et al. 2001) became the first targeted therapy introduced against those two seemingly different diseases, with remarkable efficacy and tolerability that altered their natural history significantly.

As more than a decade has passed since the introduction of imatinib into clinical practice and patients with CML or GIST may have received imatinib for many years, it is important to consider the possible long-term consequences of imatinib therapy. The scope of this paper is to review the safety of imatinib, both short and long-term, in CML and GIST.

Imatinib efficacy in CML

CML is a clonal disorder of haematopoiesis that generally progresses through three disease phases: chronic stable phase (CP), accelerated phase (AP) or blast crisis (BC). The median age at diagnosis for CML is 66 years of age (Sawyers 1999). Treatment options include allogeneic haematopoietic cell transplantation (HCT), which potentially can lead to cure, palliative therapy with cytotoxic agents and recently, treatment with tyrosine kinase inhibitors (imatinib, dasatinib and nilotinib) that can control the disease (Baccarani et al. 2009).

A pioneer phase I trial of imatinib in CML found that it was well tolerated and had significant antileukemic activity in patients with CP-CML after failure of interferon α (IFNα) treatment (Druker et al. 2001; Peng et al. 2004). Subsequent phase II studies confirmed the marked effect of imatinib as a second-line treatment in CP-CML, both after IFNα failure (Kantarjian et al. 2002) and in AC- and BC-CML (Sawyers et al. 2002; Talpaz et al. 2002). Thus, the 400 mg dose became the ‘standard dose’ for CP-CML (Peng et al. 2004) and the 600 mg dose for accelerated phase- (Talpaz et al. 2002) and BC-CML (Sawyers et al. 2002). Based on clinical results from the IRIS study (International Randomized Study of IFN vs. STI571), imatinib was approved for first-line treatment of all phases of CML (O’Brien et al. 2003). The IRIS study established the superiority of imatinib compared with IFNα and low-dose cytarabine combinations with regards to the rate of haematologic response (HR), cytogenetic response (CgR), and molecular response (MolR)(O’Brien et al. 2003), as well as in patients, who crossed over from interferonα plus cytarabine to imatinib (Guilhot et al. 2009). Moreover, preliminary results of a phase III study (TOPS), which evaluated the treatment with imatinib 800 mg versus 400 mg in patients with newly diagnosed CP-CML,showed similar MMR rates at 1 year, but both MMR and CCyR occurred earlier in patients treated with high-dose imatinib (Cortes et al. 2010). However, since the introduction of more potent second-generation TKIs (e.g. dasatinib and nilotinib), which appear to be more effective compared to imatinib, they may become the treatment of choice for patients with advanced disease (Cortes et al. 2008).

Imatinib efficacy in GIST

GISTs are mesenchymal neoplasms that appear to arise from the interstitial cells of Cajal of the myenteric plexus; median age at presentation is 60 years (Rubin et al. 2000). More than 10 years ago specific mutations of the KIT proto-oncogene that result in the constitutive activation of KIT signalling were described (Hirota et al. 1998), and increased expression of KIT protein (CD117) became a key diagnostic parameter for GIST (Kindblom et al. 1998). After the promising results in CML, imatinib, which also inhibits KIT, was administered to a patient with metastatic GIST (Joensuu et al. 2001). The patient had a rapid, substantial and durable response (Joensuu et al. 2001). Subsequently, a phase I EORTC study of imatinib in GIST was performed, in which, 36 patients received imatinib at doses ranging from 400 to 1,000 mg daily (van Oosterom et al. 2001). This study concluded that the maximum tolerated dose (MTD) of imatinib was 400 mg twice daily; imatinib had significant activity in advanced GIST with clinical benefit in 80% of patients after 9 months of therapy (van Oosterom et al. 2001). Simultaneously, the randomised B2222 phase II study was performed in North America, which compared 400 or 600 mg of imatinib daily in patients with advanced GIST (Demetri et al. 2002). This study demonstrated similar activity with an estimated 1-year survival rate of 88% (Blanke et al. 2008b) as a result of which the FDA approved imatinib for the treatment of locally advanced or metastatic GIST in February 2001 (Dagher et al. 2002). A phase II extension of the EORTC phase I trial (Verweij et al. 2003) and a Japanese phase II study (Nishida et al. 2008) confirmed the findings of the US trial. The long-term results of the B2222 study showed an overall response rate (RR) of 68.1% with nearly 50% of patients either achieving partial response (PR) or stable disease (SD) and a median survival of approximately 5 years (Blanke et al. 2008a).

Subsequently, two randomized phase III studies conducted almost simultaneously assessed the efficacy of imatinib at two dose levels (400 vs. 800 mg daily); the EORTC trial (62005) (Verweij et al. 2004) and the North American Sarcoma Intergroup study (S0033) (Blanke et al. 2008b). The GIST meta-analysis group presented the results of these two randomised studies in 1,640 patients with a median follow-up of 45 months ((MetaGIST) 2010). OS was identical in the two arms, with a small but significant PFS advantage in the high-dose arm. The multivariate prognostic models confirmed as adverse factors are as follows: male sex, poor PS, high baseline neutrophils, advanced age, large tumour size and low albumin level. Wild-type KIT, exon 9 and other mutations were correlated with poorer prognosis compared to exon 11 mutations. Interestingly, the influence of dose–effect on PFS and prognostic factors was significant only for exon 9 mutations; patients with exon 9 mutation tumours had a longer PFS when treated with 800 mg rather than 400 mg ((MetaGIST) 2010).

Most importantly, the use of imatinib also has helped to unlock the molecular mechanisms underlying growth stimulation and inhibition of cell death in GIST. We now have a better understanding of the complexity of signalling in terms of the prognostic impact of different mutations in KIT (Debiec-Rychter et al. 2006), the existence of rarer mutations in platelet-derived growth factor receptor α (PDGFRα) (Heinrich et al. 2003, 2006), the types of secondary mutations that confer resistance to imatinib (Heinrich et al. 2006) and alternative signalling pathways that may drive the disease (Janeway et al. 2007).

Adjuvant imatinib has recently been licensed in Europe for the treatment of patients after complete resection, who are at significant risk of relapse, following a randomised phase III clinical trial (ACOSOG Z9001), in which, patients were randomised to receive imatinib at 400 mg daily or placebo for 12 months; this trial showed a highly significant benefit in PFS (Dematteo et al. 2009). Currently, there are various issues concerning adjuvant imatinib (Gronchi et al. 2009), such as the possibility that occult resistant disease might develop during exposure to imatinib in the adjuvant setting. Hopefully, a large EORTC phase III trial that randomised patients with intermediate or high-risk GIST to 2 years of imatinib versus observation and the phase III trial of the Scandinavian Sarcoma Group, in which, the patients with high-risk GIST are randomized to receive imatinib for 1 or 3 years, will provide answers regarding time to imatinib failure and optimum duration of treatment (Gronchi et al. 2009).

Known toxicities

Although CML and GIST are two diseases with completely different symptomatology and natural histories, the pharmacokinetics of imatinib in GIST patients were similar to those described initially in patients with CML (Demetri et al. 2002). Thus, it is expected that there may be many similarities in the side effect profile of imatinib in these diseases.

Generally, imatinib is well tolerated overall, though mild to moderate toxicities were common in both CML and GIST with virtually all patients experiencing at least one mild or moderate adverse event (grade 1–2) that is related to therapy (Blanke et al. 2008b; Cortes et al. 2010; Dematteo et al. 2009; Demetri et al. 2002; O’Brien et al. 2003; Verweij et al. 2004). The toxicities of imatinib are determined by the disease phase as well as the dose, with more frequent and severe toxicities in patients with advanced disease (Sawyers et al. 2002; Talpaz et al. 2002) and those treated with higher doses (Blanke et al. 2008b; Cortes et al. 2010; Verweij et al. 2004). Indeed, patients with advanced disease had more dose interruptions and reductions when on high-dose treatment (Blanke et al. 2008b; Cortes et al. 2010; Verweij et al. 2004). Thus, successful treatment in these patients relies on prompt and effective management of adverse events and toxicities.

Most side effects occur early in the course of treatment (Verweij et al. 2003, 2004) and tend to reduce in frequency and intensity by time, a fact that contrasts with usual experience with conventional cytotoxic drugs (Verweij et al. 2004). A possible explanation could be enhanced drug clearance over time, which leads to a reduction in exposure (Judson et al. 2005). This could also explain the observation that patients initially treated with imatinib at 400 mg and subsequently escalated after PD at 800 mg did not experience the same degree of adverse effects as patients initially started at 800 mg daily of imatinib (Zalcberg et al. 2005). Moreover, in a GIST study, the drug clearance was lower, and subsequently, the area under the concentration curve (AUC) was higher for female patients and for patients with low baseline albumin levels (Judson et al. 2003). Those two findings suggest that drug exposure may be higher in women and/or elderly patients, who are consequently at a higher risk of toxicity (Van Glabbeke et al. 2006).

Haematologic toxicity

In CP-CML patients (IRIS trial) grade 3 or 4 haematological toxicity of imatinib 400 mg was relatively frequent (O’Brien et al. 2003) (Table 1). Despite high rates of neutropenia (14.3%) and thrombocytopenia (7.8%), grade 3–4 anaemia was observed in only 3% of patients (O’Brien et al. 2003). The respective infectious and bleeding complications remained rare. The fact that imatinib does not cause mucositis may partly account for the fact that the incidence of neutropenic fever is much less common in patients on imatinib, compared to conventional chemotherapy (Mauro and Deininger 2009).
Table 1

Summary of the Imatinib-related haematologic and non-haematologic adverse effects in phase III studies in CML and GIST

Adverse events grade 3–4. Overall (%)

CML

GIST

Newly diagnosed in chronic phase

Advanced or metastatic

Adjuvant

Phase III trials

IRIS IM versus IFNα and Ara-C

IRIS IM after IFNα and Ara-C

TOPS IM 400 versus 800 mg/d

EORTC 62005 IM 400 versus 800 mg/d

S0033, IM 400 versus 800 mg/d

Z9001 IM versus placebo

n = 551

n = 359

n = 157

n = 319

n = 470

n = 472

n = 553

n = 553

n = 337

400 mg

400 mg

400 mg

800 mg

400 mg

800 mg

400 mg

800 mg

400 mg

 

2001

2006

        

Haematologic

 Anaemia

3.1

4.4

5.0

3.8

7.0

7.0

16.7

9.0

14

 Neutropaenia

14.3

16.7

26.2

18.5

28.5

7.0

7.0

7.0

10

3.6

 Thrombocytopenia

7.8

8.9

11.1

10.2

18.0

1.5

1.3

 Haemorrhage

0.7

1.8

3.1

2.7

8.0

6.0

11

Non-haematologic

Gastrointestinal

 Nausea/vomiting

0.7/1.5

1.3/2.0

0.3/0.3

0/0.6

1.3/0.9

2.5/2.7

3.2/3.0

9.0

16

2.0/2.0

 Diarrhoea

1.8

3.3

2.2

0

2.5

1.4

5.3

9.0

16

2.0

 Fluid retention (oedema)

0.9

2.5

1.7

0

2.7

2.8

9.1

2.0

 Cutaneous (rash, dermatitis)

2.0

2.9

1.7

2.5

5.7

2.3

5.3

4.5

7.5

3.0

Musculoskeletal

 Muscle cramps

1.3

2.2

1.4

0.6

0.9

 Arthalgias—myalgias

2.4/1.5

2.5/1.5

2.2/1.7

0.6/0

1.9/3.5

0/0.2

0.8/1.1

 Muscluloskeletal pains

2.7

5.4

3.3

0

2.2

11

12

 Cardiovascular

 

7 (general)

14

1.0 (syncope)

Miscellaneous

 Fatigue

1.1

1.8

2.2

1.9

1.6

5.9

10.8

1.5

 Headache

0.4

0.5

1.4

0

1.6

0.2

0.8

 Dizziness

0.9

0

0.9

0.2

0.4

 Cough

0.2

0.2

0.3

0.2

 Anorexia

0

1.9

1.7

Laboratory abnormalities

 Transaminitis

5.1

5.3

4.7

4.5

2.8

4.0

3.5

2.5

The haematologic toxicity of imatinib treatment in CML patients does not only reflect the direct toxic effects of imatinib on stem cells due to the inhibition of KIT, but may also reflect the efficient elimination of the BCR-ABL-positive leukaemic cells and the subsequent delayed restoration of normal haematopoiesis. Thus, the reported myelosuppression could be a mixture of ‘response effect’ and imatinib-related side effect (Demetri et al. 2002; Mauro and Deininger 2009).

As expected, higher doses of imatinib (600–800 mg daily) for early CP patients resulted in higher rates of all-grade and grades 3–4 haematologic adverse events in several single-armed studies (Hughes et al. 2008) and in the phase III study (TOPS) that compared imatinib 400 mg versus 800 mg in newly diagnosed, previously untreated CP-CML (Cortes et al. 2010). Moreover, myelotoxicity is generally an early event, more commonly observed during the first 12 months of treatment with imatinib (Hochhaus et al. 2008); in a phase 2 single-arm trial of imatinib in CP-CML after failure of IFNα after 12 months of treatment, neutropenia decreased from 33 to 13%, thrombocytopenia from 18 to 1.9% and anaemia from 6 to 1.4% (Hochhaus et al. 2008). The reduction of incidence of myelotoxicity over time could be attributed to: the dropout of high-risk patients with PD (Hochhaus et al. 2008), an increase of drug clearance, which subsequently may lead to a reduction in exposure (Judson et al. 2005) or largely due to the fact that there is a higher disease burden in the bone marrow at the beginning of treatment and when normal marrow function is restored the incidence of imatinib-associated myelosuppression decreases (Hochhaus et al. 2008).

Interestingly, in CP-CML patients pretreated with IFNα that were subsequently treated with imatinib, grade 3–4 neutropenia occurred in 36%, thrombocytopenia in 22% and anaemia in 8% (Hochhaus et al. 2008), which is double the rate of grade 3/4 cytopenias compared to that observed in the IRIS trial (O’Brien et al. 2003). More frequent grade 3–4 haematological toxicity has also been observed in AP patients compared with CP-CML patients; grade 3–4 neutropenia was observed in 60% of CP patients receiving 600 mg once daily, while severe anaemia was observed in 47% and thrombocytopenia in 43% of AP patients (Talpaz et al. 2002). Similarly, in BC-CML grade 3–4, neutropenia was observed in 64% of patients, thrombocytopenia in 62% and anaemia in 52% of patients. Febrile episodes in BC patients receiving imatinib were more frequent compared with the less advanced phases of the disease (Sawyers et al. 2002).

The higher frequency of haematological toxicity in patients pretreated with IFNα and in advanced phases of CML can be attributed either to the more advanced phase of the disease, including greater bone marrow involvement, or to the higher dose of imatinib (600 mg) that is used in the treatment of advanced disease. The fact that myelosuppression is correlated with the disease phase could represent an impact of the leukaemic process on normal stem cell recruitment into cell cycle, or maturation or simply a mass effect of leukaemic blasts crowding out normal haematopoiesis (Van Glabbeke et al. 2006). Indeed, BC patients’ haematological adverse effects were difficult to distinguish from the initial grade of myelosuppression due to the underlying leukaemia (Van Glabbeke et al. 2006). Thus, as myelosuppression may reflect treatment efficacy, it is suggested not to withdraw treatment or reduce the dose of imatinib; continuation of imatinib despite myelotoxicity may be desirable in some patients (Van Glabbeke et al. 2006).

The toxicity profile of imatinib in the treatment of GIST has been evaluated in two phase III trials that compared imatinib 400 with 800 mg (Blanke et al. 2008b; Verweij et al. 2004) (Table 1). The most common grade 3–4 haematological toxicities observed in the EORTC phase III trial were anaemia and granulocytopenia (Verweij et al. 2004), while in the S0033 phase III trial, about 20% of patients on imatinib 400 mg and 27% of patients on 800 mg experienced > grade 3 haematologic toxicity (Blanke et al. 2008b). As expected, patients treated with high-dose imatinib experienced more frequent and more severe haematological toxicities. Interestingly, patients who were initially treated with 400 mg and then crossed over to imatinib 800 mg experienced more severe anaemia, which subsequently stabilized (Zalcberg et al. 2005). In contrast, neutropenia was less acute after crossover (Zalcberg et al. 2005). Most side effects happened early in the course of treatment, similar to observations in previous phase II GIST and CML studies (Hochhaus et al. 2008; Verweij et al. 2003).

Overall, imatinib is well tolerated in both patients with GIST as well as with CML (Blanke et al. 2008b; Cortes et al. 2010; Dematteo et al. 2009; Demetri et al. 2002; O’Brien et al. 2003; Verweij et al. 2004). Myelotoxicity in terms of leukopenia, neutropenia and thrombocytopenia was less striking in patients with GIST; according to phase III studies, severe neutropenia (grade 3–4) was observed in 7% in advanced GIST and 3.6% in the adjuvant setting in patients with resected localised GIST, compared to 15% with 400 mg dose and 27% with 800 mg dose in CP-CML, respectively (Table 1). This must be explained by the impact of CML on bone marrow function. Of note, although thrombocytopenia is more common in CML patients, serious gastrointestinal (GI) haemorrhages are more often seen in GIST patients (5–11% depending on dose) (Blanke et al. 2008b; Demetri et al. 2002). These haemorrhages could be related either to the underlying disease or to the tumour degeneration induced by imatinib (Demetri et al. 2002).

Interestingly, there is a marked difference in the incidence of anaemia between GIST and CML; grade 3–4 anaemia in GIST phase III trials was 7% with 400 mg and 7–10% with 800 mg (Blanke et al. 2008b; Verweij et al. 2004), while in CML phase III studies, only 3.5% of patients on 400 mg and 6% on 800 mg had severe anaemia (Cortes et al. 2010; O’Brien et al. 2003) (Table 1). Possibly, anaemia is an imatinib specific side effect and not only part of the clinical features of myelotoxicity. In support of this are the following observations:
  1. 1.

    anaemia is a dose-related toxicity of imatinib in GIST patients, while neutropenia is completely dose-independent (Van Glabbeke et al. 2006)

     
  2. 2.

    after cross-over of GIST patients from 400 mg dose to high-dose imatinib, only anaemia worsened in terms of haematologic toxicity (Zalcberg et al. 2005)

     
  3. 3.

    the initial low haemoglobin levels were found to be an independent prognostic factor for both anaemia and neutropenia, and the AUC for imatinib was higher in patients with baseline low haemoglobin level (Kantarjian et al. 2007), which translate to higher drug exposure (Van Glabbeke et al. 2006)

     
  4. 4.

    haemoglobin may play a significant role in imatinib transport and delivery (Guetens et al. 2003).

     

Non-haematologic toxicity

Gastrointestinal toxicity (GI), presenting either as nausea and vomiting or diarrhoea, is among the most common side effects of imatinib in both CML and GIST patients. In CML, nausea is the most common GI side effect occurring in 43–65% patients taking 400–600 mg daily of imatinib (Mauro and Deininger 2009). However, severe (grade 3–4) nausea occurs in less than 1% of cases (Cortes et al. 2010; O’Brien et al. 2003). Diarrhoea is also common, occurring in 45% of imatinib-treated patients, while severe diarrhoea ranges between 1.8 and 2.5% (Mauro and Deininger 2009).

In GIST patients, grade 1–2 GI events (nausea, vomiting, diarrhoea and flatulence) were common. In a phase I study, nausea and vomiting were among the toxicities that determined the MTD of imatinib to be 800 mg (van Oosterom et al. 2001). In the phase II (Demetri et al. 2002; Nishida et al. 2008; Verweij et al. 2003) and III studies that followed, both in advanced (Blanke et al. 2008b; Verweij et al. 2004) and adjuvant settings (Dematteo et al. 2009), nausea was observed in 52–57% and diarrhoea in 45–52%. Severe toxicities were fewer; grade 3–4 nausea–vomiting in 3% and diarrhoea in 1.5–5.3% of patients (Blanke et al. 2008b; Dematteo et al. 2009; Verweij et al. 2004). Expectedly, nausea and diarrhoea have been observed more frequently in patients receiving high-dose imatinib (Blanke et al. 2008b; Dematteo et al. 2009; Verweij et al. 2004). Nausea was one of the main reasons that led to dose reductions in patients receiving the 800 mg dose (Blanke et al. 2008b). Moreover, high dose, female sex, poor PS (for nausea only) and GI site (for diarrhoea only) were identified as independent prognostic factors in univariate and multivariate prognostic risk factor analyses (Van Glabbeke et al. 2006). Although there are no randomised data for CML, most studies also suggest that these toxicities are dose dependent (Van Glabbeke et al. 2006).

Severe nausea, vomiting and diarrhoea were more often seen in GIST than in CML patients, when treated with the same dose of imatinib. This may be due to the differences in the natural history of both diseases, as GIST may cause GI symptoms, including nausea and diarrhoea. Moreover, patients with GIST more often undergo GI surgery. Since this most frequently involves a fairly major gastric resection, this may cause dumping syndrome, which can further complicate the case. In support of the above is the fact that the GI site is an independent prognostic factor for diarrhoea (Van Glabbeke et al. 2006).

Fluid retention is the most typical non-haematological side effect of imatinib for both CML and GIST and one of its dose limiting toxicities and main reasons for dose reductions of high-dose imatinib (Blanke et al. 2008b; Verweij et al. 2004). The majority of patients have some degree of superficial oedema (Blanke et al. 2008b; Druker et al. 2006; Zalcberg et al. 2005), which is usually mild (grade 1–2), dose dependent and often present in the periorbital regions and ankles. The imatinib-induced fluid retention syndrome is a side effect sui generis. It has been speculated that the syndrome is caused by imatinib inhibition of platelet-derived growth factor receptor (PDGFR), which plays an important role in vasculogenesis (Kantarjian et al. 2007), or imatinib inhibition of KIT that is expressed in dermal dendrocytes, which are bone marrow-derived cells (Esmaeli et al. 2002).

Interestingly, it is most commonly reported in patients with advanced GIST. Specifically, in phase II (Demetri et al. 2002; Verweij et al. 2003) and phase III trials (Blanke et al. 2008b; Verweij et al. 2004) of imatinib administered in patients with advanced GIST oedema was noted in 74–84%, while in CP-CML studies, it was noted in 54–65% (Cortes et al. 2010; O’Brien et al. 2003). Respectively, as shown in Table 1, severe fluid retention resulting in pleural or pericardial effusions or even oedema anasarca was rare but most often observed in GIST patients (2.8–9% vs. 1–1.7% in CP CML). In CML, oedema is more common in advanced phases (up to 3%), which reflects not only the advanced stage of CML and associated co-morbidities but also the higher dose used in these stages (Sawyers et al. 2002). Independent prognostic factors for oedema in GIST included high dose, female sex and baseline low albumin level (Van Glabbeke et al. 2006). Thus, a possible explanation of the different incidence of oedema between these two diseases could be the low baseline albumin levels that are more often observed in advanced GIST; there is no study directly comparing the toxicity of imatinib in GIST and CML.

Skin toxicity, mostly as a rash or dermatitis, is frequently reported (Mauro and Deininger 2009). It is typically mild to moderate, usually appearing soon after commencement of therapy. All grades of rash have been reported; with an incidence of 40% in CML patients (Cortes et al. 2010; O’Brien et al. 2003) and 30–70% in GIST patients (Blanke et al. 2008b; Verweij et al. 2004). Higher grade reactions are rare with an incidence of 2–3% with 400 mg of imatinib and 5–7% with the 800 mg dose (Blanke et al. 2008b; Cortes et al. 2010; O’Brien et al. 2003; Verweij et al. 2004). Severe dermatologic reactions, including erythema multiform and Stevens–Johnson syndrome, have been reported in both diseases in individual case studies (Scheinfeld 2006).

As shown in GIST patients, the development of rash is dose dependent and more frequent in advanced age. Surprisingly, small size of tumour lesions was also an independent prognostic factor (Van Glabbeke et al. 2006). Though in theory this correlation could be attributed to the fact that patients with large lesions tend to progress earlier, this hypothesis was rejected, as tumour size remained a significant risk (Van Glabbeke et al. 2006). In the CML studies, a higher risk of toxicity has been reported for patients with advanced disease, but the higher dose of imatinib that is usually used in these cases may be an important confounding factor (Sawyers et al. 2002; Talpaz et al. 2002). Until now, there is no explanation for this correlation.

Other commonly observed non-haematologic toxicities in patients suffering from CML and GIST on imatinib treatment include fatigue, musculoskeletal toxicity—cramps, myalgias and arthalgias-headache, dizziness, cough and anorexia (Table 1). Severe toxicity (grade 3 or 4) has been observed more frequently in patients receiving imatinib 800 mg daily (Blanke et al. 2008b; Cortes et al. 2010). Concerning fatigue, advanced age and poor PS were also independent prognostic factors (Van Glabbeke et al. 2006); moreover, fatigue was one of the main reasons for dose reductions in patients receiving high dose imatinib (Blanke et al. 2008b), and was aggrevated when patients crossed over from 400 mg to 800 mg (Sawyers et al. 2002); like anaemia, this was attributed both to disease progression and drug specific toxicity.

Since liver is the most common site of GIST metastases (60%) (Langer et al. 2003) and imatinib is metabolized by hepatic CYP enzymes (Ramanathan et al. 2008), it might have been expected that hepatotoxicity (transaminitis) would be more severe in advanced GIST compared to CML. However, the incidence of mild transaminitis because of imatinib treatment is similar for CML and GIST patients, and severe transaminitis occurred in only 2.5–5% of patients (Blanke et al. 2008b; Cortes et al. 2010; O’Brien et al. 2003). This can be explained by the fact that the plasma pharmacokinetics and urinary excretion of imatinib did not differ between patients with normal liver function and those with varying degrees of liver dysfunction, and that there is no correlation between the imatinib pharmacokinetics and LFTs or the occurrence of DLTs (Ramanathan et al. 2008). Moreover, it has been shown that in advanced GIST with liver metastases, there was no difference either in the effectiveness of imatinib or the incidence of imatinib adverse events (Zhu et al. 2010).

Long-term side effects

With the advent of effective signalling inhibitors, such as imatinib, the concept of the need for continuous treatment has become more or less accepted. This was already acknowledged to be advisable for chemical castration treatment of prostate cancer, and now, CML and GIST have been added to the list of malignancies, for which continuous treatment is necessary. Current thinking suggests that imatinib therapy must not be discontinued in patients with CML remission (Mauro and Deininger 2009), as imatinib interruption resulted in a rapid molecular relapse in 50% of patients in a phase II trial (Rousselot et al. 2007). Only recently it was published that imatinib can be safely discontinued in patients with a complete MolR of at least 2 years duration (Mahon et al. 2010). Imatinib interruption in the setting of advanced GIST results in rapid progression in most patients (Le Cesne et al. 2010) and is not recommended outside clinical trials except in case of intolerable side effects. In the adjuvant setting for GIST, it is unknown whether short duration treatment will exert a prolonged benefit, though we await the results of two randomized trials that might shed light on this question.

Hence, as more patients undergo prolonged treatment with imatinib and may potentially be cured, it is necessary to focus on potential long-term side effects of imatinib. Fortunately, long-term follow-up of patients on imatinib in both diseases demonstrates excellent tolerability. Specifically, the 6-year follow-up of patients with CP-CML on imatinib showed no additional serious safety issues, and there was no increase in serious laboratory abnormalities over time (Hochhaus et al. 2008). Similarly, the ILTE (Imatinib Long-Term Effects) study confirmed that CML patients on imatinib do not show new types of imatinib-related adverse events after 8 years of follow-up, do not appear to have substantially higher second cancer rates than the general population, and have mortality rates lower than expected in an age-/sex-matched population (Gambacorti-Passerini et al. 2008). Respectively, imatinib was also well tolerated over long-term administration in patients with metastatic GIST, as no new serious adverse events emerged and no patient withdrew from treatment because of adverse events (Blanke et al. 2008a).

Cardiotoxicity

In an early report, imatinib was associated with the development of severe heart failure (Kerkela et al. 2006), which alarmed the oncology community and prompted the manufacturer to revise imatinib labelling regarding possible heart failure (HF) and left ventricular dysfunction (LVD). Laboratory studies are conflicting; in one study, imatinib was found to be cardioprotective in hypertensive rats; it attenuated the decline in cardiac function and reduced renal microvascular damage both in animals and in cell cultures (Schellings et al. 2006). In contrast, other studies indicated a cardiotoxic effect of imatinib mediated by inhibition of ABL, which was depicted clinically as severe LVD and HF in 10 patients (Kerkela et al. 2006), or by describing a structurally reengineered form of imatinib that inhibited KIT but not ABL and showed less cardiotoxicity (Fernandez et al. 2007). Another study regarded imatinib as a potential cardiotoxin, as it may affect cardiac AKT and ERK 1/2 phosphorylation that predisposes to cardiac failure under conditions of stress (Trent et al. 2010).

In order to identify the clinical impact of imatinib’s cardiotoxicity, subsequent evaluation of the incidence of HF among haematological and GIST patients receiving imatinib on clinical trials showed that imatinib is a very uncommon cause of HF or LVD, observed at rates similar to those expected in the general population (Dematteo et al. 2009; Trent et al. 2010; Verweij et al. 2007). Specifically the calculated incidence of cardiotoxicity in CML and GIST patients on imatinib treatment was estimated as 0.2% per year (Hatfield et al. 2007; Verweij et al. 2007). Moreover, it was suggested that HF in patients with underlying cardiovascular disease could have been exacerbated by imatinib-associated fluid retention, a known common side effect of imatinib (Atallah et al. 2007; Hochhaus et al. 2008; Verweij et al. 2007). The weakness of these studies is that are retrospective in nature and have not relied on prospectively defined measurements of cardiac function; in addition the known overlap of HF symptoms, like dyspnoea, fatigue, and oedema, with imatinib side effects makes clinical assessment difficult. Thus, it is possible that the reported incidence of imatinib induced HF may be underestimated.

Lately, it was found that cytotoxic concentrations of imatinib were required to trigger markers of apoptosis and stress response in neonatal rat ventricular myocytes and fibroblasts, and it was suggested that imatinib is not cardiotoxic at clinical concentrations (Wolf et al. 2010). A recent study that prospectively evaluated CML patients treated with imatinib for cardiotoxicity, using echocardiography and MUGA scanning, reported no evidence of myocardial deterioration over 12 months of imatinib treatment (Estabragh et al. 2011), while in another, in which, patients with GIST receiving imatinib were monitored with serum levels of brain natriuretic peptide (BNP), substantial increases in BNP were seen in 4% of patients after a 3-month period, suggestive of possible subclinical HF (Perik et al. 2008).

Thus, more prospective studies with objective cardiac monitoring are warranted to determine the incidence and clinical significance of imatinib-induced HF. Until this issue is clarified, although there is no indication for routine cardiac monitoring of patients treated with imatinib, it is recommended in clinical practice to consider patients on imatinib treatment as stage A HF patients, i.e. at risk for HF, but without structural heart disease or symptoms (Trent et al. 2010). Management of such patients involves modification of risk factors that may predispose to HF, including treatment of blood pressure and hypercholesterolemia (Trent et al. 2010). Certainly, patients with significant cardiac history should be followed closely and symptoms suggestive of systolic heart failure should be properly investigated and managed when they occur to avoid the need for dose reduction of imatinib.

Bone metabolism

Changes in bone parameters may occur in imatinib-treated patients (Berman et al. 2006; Grey et al. 2006). In one preliminary report, patients receiving imatinib had increases in markers of bone formation (serum osteocalcin and P1NP) but no change in a marker of bone resorption (beta isomer of C-terminal telopeptide of type I collagen) within 3 months of initiating therapy (Grey et al. 2006). Moreover, hypophosphatemia has been reported during treatment, along with low–normal serum calcium and secondary hyperparathyroidism (Berman et al. 2006; Grey et al. 2006; Osorio et al. 2007).

Subsequent studies in CML patients receiving imatinib showed that imatinib significantly increased bone mineral density (BMD) (Fitter et al. 2008; Jonsson et al. 2008; O’Sullivan et al. 2009). Specifically, in one study, a twofold increase in trabecular bone volume (TBV) in iliac crest bone was shown after 2–4 years of imatinib therapy (Fitter et al. 2008). Another study showed higher BMD in the lumbar spine usually higher than would be expected for age and gender (Jonsson et al. 2008).

This probably results from inhibition of osteoclasts and activation of osteoblast activity through inhibition of M-CSFR, KIT and PDGFR signalling (Vandyke et al. 2010). Specifically, it is proposed that the proliferation and survival of osteoclast precursors and mature osteoclasts are driven by M-CSFR signalling (Dewar et al. 2005). In vitro studies demonstrated that the inhibition of M-CSFR, KIT and PDGFR signalling by imatinib inhibits osteoclastogenesis not only in terms of activity but also their numbers as it increases apoptosis of osteoclasts (Dewar et al. 2006; El Hajj Dib et al. 2006; O’Sullivan et al. 2009).

Simultaneously, the proliferation and differentiation of osteoblasts is affected, as it is regulated among other signalling pathways, through ABL and PDGFR pathways (Vandyke et al. 2010). Inhibition of ABL is thought to inhibit osteoblast activity (Li et al. 2000), while inhibition of PDGFR signalling increases osteoblast differentiation and simultaneously inhibits the production of osteoclastogenic cytokines by osteoblasts (O’Sullivan et al. 2007).

These changes in bone remodelling parameters would theoretically result in a decrease in the dissolution of calcium and phosphate from bone or in an increase in the sequestration of calcium and phosphate in newly formed bone, resulting in hypophosphatemia and hypocalcaemia (Vandyke et al. 2010). Decreased serum calcium causes increased PTH production and secondary hyperparathyroidism, which in turn increases calcium reabsorption and phosphate excretion by the kidney (O’Sullivan et al. 2009; Vandyke et al. 2010). Collectively, imatinib modulates bone homeostasis, and these proposed mechanisms explain the observed effects on markers of bone turnover and bone density.

The long-term consequences of dysregulated bone remodelling by imatinib are not known. Although imatinib may increase bone strength, through positive effects on BMD and TBV, there are concerns, derived from studies on biphosphonates (Vandyke et al. 2010), that long-term inhibition of bone turnover may eventually result in weakened bone through the accumulation of microfractures, and, subsequently lead to decreased bone strength (Komatsubara et al. 2003). Moreover, there may be other possible unanticipated long-term side effect of imatinib; an example could be long-term effects on haematopoiesis indirectly via the impact of imatinib on osteoblast activity, as they are considered vital components of the hematopoietic stem cell niche (Chitteti et al. 2010; Wu et al. 2009).

Further research is warranted as the clinical consequences both short- and long-term are not entirely known. Currently, there are no specific monitoring recommendations for changes in bone density in patients treated with imatinib. Based on the median age of CML and GIST patients, most of them are eligible for osteoporosis monitoring; it can also be suggested to include periodic electrolyte studies, in order to identify early cases, in which a further workup may be required.

Glucose metabolism: diabetes

In vitro studies have shown evidence that the control of glucose metabolism could be an important mechanism of the antiproliferative action of imatinib (Gottschalk et al. 2004). Another study showed that imatinib markedly reduces glucose uptake in GIST cells via decreased levels of plasma membrane bound Glut4 in an AKT-dependent manner, while the inhibition of cell growth and induction of apoptosis is independent of glucose deprivation and mediated through an AKT-independent manner (Tarn et al. 2006). These findings suggest that imatinib-mediated decrease in tumour metabolism observed on positron emission tomography (PET) using fluorine-18-fluorodeoxyglucose (18FDG) (Van den Abbeele 2008) may not entirely reflect imatinib efficacy. In other words, the loss of FDG uptake on PET scan may be a surrogate for imatinib activity but not a direct measure of antitumour activity (Tarn et al. 2006).

Subsequent studies confirmed the in vitro findings; the fasting blood glucose in diabetic CML patients that achieved a complete CgR improved after 3 months of treatment with imatinib (Breccia et al. 2004; Veneri et al. 2005), as well as in GIST patients (Agostino et al. 2010), while symptomatic hypoglycaemia was observed in non-diabetic patients with advanced GIST treated with imatinib (Hamberg et al. 2006). However, another retrospective study demonstrated that imatinib did not modify substantially the glycemic profile in patients with CML and GIST (Dingli et al. 2007).

Overall, the molecular mechanisms of the effects of imatinib on glucose metabolism are still unknown, but several hypotheses have been made. According to one, imatinib prevents beta-cell apoptosis and maintains the pancreas insulin secretion capacity in situations of beta cells stress, through promotion of NF-kB activation (Hagerkvist et al. 2007). In a retrospective study, the hypoglycaemic effect of imatinib was attributed possibly to KIT and/or PDGFRβ inhibition (Agostino et al. 2010).

Recently, plasma adiponectin levels in CML patients were found elevated threefold after 3 months of imatinib treatment (Fitter et al. 2010). Adiponectin is a hormone that is considered to play an important role in the regulation of insulin sensitivity, as it increases glucose uptake and fatty acid oxidation in muscle and reduces hepatic gluconeogenesis (Yamauchi et al. 2002); indeed, adiponectin levels increase before an improvement in insulin sensitivity is observed (Yang et al. 2001). The mechanism by which plasma adiponectin levels increase after imatinib therapy has not yet been fully elucidated; according to an in vitro study, imatinib may possibly promote adipocyte differentiation of bone marrow-derived mesenchymal stromal cells (Fitter et al. 2008).

Conclusion

In summary, imatinib has a good safety profile and is very effective in the treatment of CML and GIST and provides a durable long-term therapy. It is currently the approved first-line TKI for newly diagnosed CP-CML in chronic phase and of unresectable, metastatic or surgically incompletely resected GIST, as well as in the adjuvant setting of high risk, completely resected GIST. There are very few significant concerns regarding long-term side effects, and those that have emerged, like cardiotoxicity, have thus far turned out to be exaggerated.

Acknowledgments

Eirini Thanopoulou is a scholarship recipient of the Hellenic Society of Medical Oncologists (HeSMO).

Copyright information

© Springer-Verlag 2011