An open-labelled study of granulocyte colony-stimulating factor in the treatment of active Crohn's disease


  • J. R. Korzenik,

    1. Department of Internal Medicine, Inflammatory Bowel Disease Center, Gastrointestinal Unit, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA
    Search for more papers by this author
  • B. K. Dieckgraefe

    1. Department of Internal Medicine, Division of Gastroenterology, Washington University School of Medicine, St Louis, MO, USA
    Search for more papers by this author

Dr J. R. Korzenik, Gastrointestinal Unit, Massachusetts General Hospital, Blake 4, 55 Fruit Street, Boston, MA 02114, USA.


Background : Immunodeficiency syndromes associated with a Crohn's-like illness suggest innate immune defects may lead to Crohn's disease. Anecdotal cases using haemopoietic colony-stimulating factors report improvement in intestinal disease associated with these syndromes.

Aim : To test the safety and efficacy of recombinant human granulocyte colony-stimulating factor in active Crohn's disease.

Methods : In an open-labelled 12-week trial, patients with a Crohn's Disease Activity Index between 220 and 450 were treated with recombinant human granulocyte colony-stimulating factor (filgrastim, Neupogen). Concomitant immunosuppressants were prohibited except prednisone ≤20 mg/day. Patient's received recombinant human granulocyte colony-stimulating factor 300 mcg daily subcutaneously adjusted to achieve an absolute neutrophil count between 25 and 35 × 109/L.

Results : Twenty patients were enrolled with a mean initial Crohn's Disease Activity Index of 307 (range: 234–428). Fifteen patients (75%) completed 8 weeks; 13 patients (65%) completed 12 weeks with the mean Crohn's Disease Activity Index for patients continuing through those times of 196 (range: 36–343) and 162 (range: 20–308), respectively. At week 12, 11 patients (55%) demonstrated a decrease of at least 70 points; five (25%) achieved a sustained remission. The mean decrease was statistically significant at each assessment time-point. Three of four patients with fistulae had a positive response. Adverse effects included bone pain, mostly mild resolving with continued treatment. One patient was hospitalized with a viral-like syndrome but it is uncertain if this was treatment related.

Conclusion : Recombinant human granulocyte colony-stimulating factor is safe and potentially effective therapy for active Crohn's disease.


While the fundamental aetiology of Crohn's disease (CD) remains unknown, the prevailing hypothesis contends that CD results from an aberrant, T cell-driven process activated by unknown stimuli. Most current and investigational therapies for CD aim to suppress the immune system to restore health. Established medical strategies for treatment of the intestinal inflammation of CD include broadly acting immunosuppressive agents such as corticosteroids, methotrexate, azathioprine or mercaptopurine (6-mercaptopurine). Recent therapeutic innovations, such as infliximab, have been designed to be more selective, interfering with specific elements of the inflammatory cascade.

An alternate understanding is that the characteristic intestinal inflammation represents a secondary response to a primary deficiency of innate immunity. This concept is suggested by the association of CD with genetic diseases, such as chronic granulomatous disease (CGD), glycogen storage disease Ib (GSDIb), and cyclic neutropenia, among others, in which a variety of qualitative or quantitative neutrophil deficiencies or other defects in innate immunity have been well characterized.1–18 These associations infer that a Crohn's phenotype may be initiated by a deficiency in neutrophil or monocyte function which, in turn, leads to chronic inflammation in which macrophage and eventually T-cell activation may occur as a sequential mucosal immune response to compensate for the deficiency in innate immunity.19

The Crohn's-like intestinal disease identified in patients with the above genetically characterized, innate immunodeficiency syndromes has been reported to respond to standard immunosuppressive medications used to treat CD.20, 21 More recently, case studies have described a therapeutic benefit of recombinant human granulocyte colony-stimulating factor (rhuG-CSF) and recombinant human granulocyte/macrophage colony-stimulating factor (rhuGM-CSF) in the intestinal manifestations of these diseases.8, 22–25 These compounds were used with the intent to increase neutrophil quantity and to augment function as well. The neutrophil deficiency syndromes and their clinical response to rhuG-CSF and rhuGM-CSF suggest an alternate model and therapeutic approach for CD using selective immune stimulation to enhance innate immunity. Based on these observations, a recent open-labelled trial of rhuGM-CSF suggested a significant benefit in CD. The current trial developed as a pilot study to examine the possible utility of G-CSF in idiopathic active CD.

In clinical practice, rhuG-CSF is widely used for treatment of neutropenic oncological patients receiving chemotherapy for malignancies.26 The rhuG-CSF promotes growth and maturation of the myeloid cell lineage and also modulates a number of neutrophil functions. Treatment with rhuG-CSF enhances neutrophil adherence, phagocytosis, superoxide anion generation and microbial killing.27–29 The rhuG-CSF also decreases apoptosis of mature neutrophils.30 While G-CSF was utilized because of its properties which augment neutrophil function, G-CSF has been suggested to have some counter-inflammatory properties as well.31 G-CSF and GM-CSF differ in important characteristics and could be acting through different mechanisms. GM-CSF acts earlier in the haematopoietic lineage and functions as a more potent stimulant of innate immunity. While G-CSF has some similar properties, it may shift a Th1 type response towards a Th2 response and acts to down-regulate tumour necrosis factor (TNF)-α among other inflammatory mediators.

We report the results of an open-labelled trial to investigate the safety and efficacy of rhuG-CSF for the treatment of patients with active CD with and without fistulous complications.


Patient selection and eligibility

In this open-labelled, 12 week trial of rhuG-CSF (filgrastim, Neupogen manufactured by Amgen, Inc., Thousand Oaks, CA, USA) for the treatment of active CD, patients were enrolled with a Crohn's Disease Activity Index (CDAI) >220 and <450. Eligible patients were initially evaluated at the Inflammatory Bowel Disease Clinic at Washington University School of Medicine. The Washington University institutional review board approved the protocol and all patients provided written informed signed consent before any study-related procedures were carried out. The study was conducted at the Washington University Center for Clinical Studies.

Patients were required to be 18 years or older with a duration of disease for at least 3 months defined endoscopically or radiologically within the previous 2 years. Negative stool studies for Clostridia difficile, ova and parasites and studies for routine pathogens were necessary before enrolment. Patients who had undergone gastrointestinal surgery in the previous 3 months, who had significant renal or hepatic disease, intra-abdominal abscess or who had clinically significant strictures were excluded. Individuals with a history of leukaemia, lymphoproliferative diseases or gout were excluded. Unexplained abnormalities in prescreening blood work were cause for exclusion as well. Contraceptive use was required by women of child-bearing potential.

Concomitant medications

Permitted concomitant medications on entry included mesalazine (mesalamine) compounds (e.g. sulfasalazine, Asacol [delayed release mesalamine, Proctor and Gamble, Cincinnati, OH, USA] and Pentasa [controlled release mesalamine, Shire Pharmaceuticals, Wayne, PA, USA]) if used for at least 8 weeks and a stable dose for 4 weeks; steroids, if on for at least 8 weeks and with a stable dose of prednisone of <20 mg/day for at least 4 weeks; antibiotics for CD if stable for 4 weeks. Patients were excluded if they had received azathioprine/mercaptopurine or methotrexate in the previous 4 weeks or infliximab within the previous 12 weeks. Non-steroidal anti-inflammatory drugs (NSAIDs) were prohibited for 10 days before enrolment and throughout the study. If steroids had been discontinued, the patients needed to have discontinued them for at least 4 weeks. A change in CD-related medication was not allowed during the study.

Design of the study

Patients were screened for eligibility within 14 days of initiating therapy. A baseline laboratory assessment included a complete metabolic panel at screening that was repeated at week 6 and at completion of the study at week 12. Erythrocyte sedimentation rate (ESR) was performed at baseline and weeks 4, 8 and 12. A complete blood count (CBC) was performed weekly throughout the study and an additional CBC was performed at midweek for the initial 2 weeks. Assessments by a doctor were performed at a screening visit (week −1) and continued at initiation (week 0), and weeks 2, 3, 4, 6, 8, 10, 12 and 16. CDAI was completed at each of the doctor visits.

Patients with primarily fistulous disease were included in a parallel protocol, which was identical except that administration of medication was discontinued at week 8 if fistulas were completely closed. If the CDAI of the patient with fistulae were >220, the patient was evaluated in both protocols.

All patients received rhuG-CSF daily for 12 weeks at an initial dose of 300 mcg self-administered subcutaneously. Patients were counselled to take acetaminophen 500 mg approximately an hour before administration of rhuG-CSF because of potential bone pain. The absolute neutrophil count (ANC) was targeted between 25 and 35 × 109/L and the dose was adjusted downward by 100 mcg if patients ANC exceeded this range. Patients whose ANC continued to be >35 × 109/L after a reduction to 200 mcg/day and a subsequent reduction to 100 mcg/day, were lowered to 75 mcg/day and continued at that dose for the remainder of the study, regardless of the ANC count.

End-points.  Safety evaluations were performed at all visits and included clinical assessment, vital signs and laboratory monitoring. Patients were monitored weekly for changes in ANC, absolute monocyte count (AMC) and platelet counts.

The primary end-point was a decrease in the CDAI of >70 points, which represents mild-to-moderate improvement. Remission was considered to be a CDAI <150. Subjects were withdrawn if their CDAI increased by 100 points over baseline on two consecutive visits or if the doctor considered the subject's condition to have worsened significantly. Statistical analyses were performed using a one-way anova to assess differences between mean CDAI scores post-treatment, and the Dunnett multiple comparisons test for individual comparisons.



Twenty patients were enrolled with active mucosal disease; three had perianal fistulas as well, one had enterocutaneous fistulas. One additional patient, not included in the evaluation, began the screening period but discontinued involvement before receiving medication because of an acute obstruction requiring surgery. Twenty patients are included in the results presented. All patients with fistulae had a CDAI >220 and are included in the mucosal disease study results. Response of fistulae to rhuG-CSF is discussed separately. Baseline characteristics and concomitant medicines are detailed in Tables 1 and 2. The mean CDAI at initiation was 307 ± 54 (range: 234–428).

Table 1.  Patient baseline characteristics
  1. CDAI, Crohn's Disease Activity Index.

Gender (M/F)3/17
Age, mean (range)38 (25–59)
Disease duration, years (range)9.3 (2–30)
CDAI on entry (range)307 (234–428)
Previous surgery7
Cigarette use4
Disease distribution
Concomitant medications
 Prednisone (previous or current use)8 (11.25 mg mean dose) (18)
 Antibiotics6 (8)
 Mesalazine10 (10)
Previous medications
Table 2.  Patient characteristics and CDAI for week 0 through week 12
NumberGenderAge (years)Duration (years)LocationCigarettesPrevious surgeryPrevious medicationsConcomitant medicationsWeek 0Week 4Week 8Week 12CDAI Δ
  1. * Patients who withdrew due to worsening disease.

  2. Patient 11 withdrew due to failure to improve. Patient 13 discontinued at week 11 due to small perianal abscess. Patient 18 withdrew choosing surgery after initiating medication. Patient 20 withdrew due to an intercurrent illness described under safety results.

  3. Gender: F = female, M = male; location: I = ileal, C = colonic, IC = ileocolonic. Cigarettes and previous surgery: Y = yes, N = no; previous and concomitant medications: SASP = sulfasalazine, mesal = mesalazine (Asacol, Pentasa), GC = glucocorticoids, cipro = ciprofloxacin, metro = metronidazole, abx = antibiotics other than ciprofloxacin and metronidazole, MP = mercaptopurine, mtx = methotrexate, inflix = infliximab.

 1F257ICNNGC, abxNone273310170109−164
 2F404CNNMP, inflixMesal, GC, Biaxin337221248169−168
 3F3215CNNInflix, MPGC, cipro. metro248688020−228
 5M507ICYYMP, abxMesal, GC428127260212−216
 6F307CNNMP, GC, abxMesal25810836163−95
 7F486CNNGC, abxMesal, GC268659972−196
 8F333CNNMP, inflixMesal241271265205−36
 9F5910INNMP, inflixNone234189299146−88
10*F405ICNYMP, abx, GCMesal285416 (week 2)   
MP, GCMesal, cipro, metro306284285  
12*F287ICYYMP, mtxGC344496 (week 2)   
13F4620ICNYMP, GC, abxMesal330159225  
14F3813INNGCMesal, metro3421336693−249
15F346ICNYGCMesal, cipro, metro288179181171−117
16M5130ICNYMP, cipro, mtxMetro, GC382272343308−74
17*F2813CNNGC, abx, MPNone248417372 (week 5)  
18F343ICYNGC, abx, MPMesal, GC270    
19F424CNNMP, inflix, GCNone364248201188−176
20F299ICYYInflix, mtxGC368    


All patients began rhuG-CSF therapy at 300 mcg dose, which provided a mean initial dose of 4.7 ± 1.0 mcg/kg (range: 6.4–2.7). Dose reductions because of an ANC greater than the target range were necessary in all but two patients. The mean final dose was 1.7 ± 0.8 mcg/kg, including one patient who received rhuG-CSF at 5.9 mcg/kg but did not require a dose decrease (range: 1.1–5.9). Responders and non-responders did not differ significantly with regard to their mcg/kg dose.


Mild transient bone pain occurred in 12 patients. Three patients experienced more severe bone pain. The administration of acetaminophen relieved the discomfort in most and with continuation of rhuG-CSF, the bone pain resolved in all patients, usually between weeks 2 and 4. One patient discontinued medication after 2 weeks because of a viral-like syndrome characterized by an unproductive cough, fever, malaise. Her CDAI had decreased slightly before discontinuation. She had an initial increase in her ANC during therapy but had a progressive decrease in her counts after week 2 without any change in dose of rhuG-CSF. She was hospitalized and an extensive evaluation failed to identify any viral, bacterial or fungal pathogen. A CD4 count was markedly low at 220, which returned to normal several weeks later.

Laboratory values

As expected, the ANC increased markedly during administration of the medication. The baseline (week 0) ANC was 6.6 ± 2.8 × 109/L with an increase to 40.1 ± 11.9 × 109/L at week 1 (range 16.8–56.3 × 109/L) and a final ANC of 27.3 ± 11.8 × 109/L at week 12 (Figure 1). Patients with the most vigorous clinical response (those with a decrease in CDAI >100 points on two visits) had a higher ANC at week 1 with a mean of 46.6 compared with non-responders mean ANC of 34.8 (P = 0.027). AMCs demonstrated an initial increase as well with a baseline count of 0.5 × 109/L at week 0, 1.3 × 109/L at week 1, and returned to baseline values by week 12. No change was seen in eosinophil count. Alkaline phosphatase increased as well with a baseline of 90 U/L (range: 56–156) to a week 1 of 270 U/L (range: 171–364) which gradually diminished in most patients during the study period with a week 12 mean of 174 U/L (range: 27–416). This was presumed due to an increase in leucocyte alkaline phosphatase. ESR showed a modest decrease from baseline 30.8 mm/h (range: 9–78) at week 8: 24.2 mm/h (range: 3–56, P = 0.005) and week 12: ESR 29.9 mm/h (range: 2–71, P = 0.06) (data set at week noted). Changes were not observed in albumin or haemoglobin.

Figure 1.

Absolute neutrophil count (109/L) by week during recombinant human granulocyte colony-stimulating factor (rhuG-CSF) treatment with S.E.M.

Clinical response

Week 4.  Sixteen of 20 patients (80%) completed the initial 4 weeks of therapy. The mean CDAI at week 4 was 229 ± 98 (range: 65–417) for all patients (last value carried forward) which represents a statistically significant change compared with week 0. The mean CDAI was 206 ± 91 (range: 65–417) for those continuing treatment at week 4 (Table 2). Ten patients (50% of the initial 20 patients) had a decrease in CDAI of least 70 points, of whom all decreased >100 points, with five patients achieving remission. Four patients withdrew: two for worsening disease, one withdrew opting to undergo surgery (CDAI week 0: 270 and week 2: 235), and one due to an intercurrent illness.

Week 8.  Fifteen patients continued through week 8; 11 (55% of the initial 20 patients) demonstrated a decrease of at least 70 points (all but one of whom decreased more than 100 points), with four achieving remission (CDAI < 150) (Table 2). The mean CDAI was 231 ± 105 for all patients which represents a statistically significant change compared with week 0. The mean for patients continuing in the study was 196 ± 86 (range: 36–343). One patient discontinued the study between weeks 4 and 8 because of worsening of disease.

Week 12.  Thirteen patients completed the 12 weeks of the study. One discontinued at week 9 due to lack of improvement. One discontinued at week 11 due to the development of a small perianal abscess at the site of a fistula, which had not been draining at the previous visit. The mean CDAI for all 20 patients was 215 ± 102 which represents a statistically significant change compared with week 0. The mean CDAI for those completing the 12-week study was 162 ± 82 (range: 20–308). Eleven had a decrease of at least 70 points and eight decreased more than 100 points. The mean decrease in CDAI among responders was 153 ± 56 (range: 88–249). Five patients were in remission (Table 2). One who had been a responder at week 8 showed an increase in CDAI with a decrease in dose due to an ANC above the target range. Another responder at week 8 discontinued at week 11 because of a small perianal abscess. Her CDAI at week 11 was 206, 124 points below her baseline score (she is not included as a week 12 responder). The overall assessment for repeated measures anova was highly significant with a P-value = 0.0026.

Week 16.  Patients were seen in follow-up 4 weeks after completion of therapy. Among the 11 responders at week 12, four maintained a response lasting an additional 4 weeks while the others had an increase in disease activity although still all below baseline (mean CDAI 81 ± 64 below week 0) (Table 2).

Fistulous disease

Three patients with perianal fistulous disease were treated with a response demonstrated in all three, defined as closure of more than 50% of fistulae. One patient had complete and prolonged closure of a solitary fistula and discontinued at week 8 as per protocol with a prolonged response of more than 6 months. A second patient had four fistulas, which had been draining for more than 10 years and had closure of all four at week 6 but had some minor drainage at week 12 in one fistula, while the others remained closed. After retreatment for an additional 12 weeks, this patient experienced complete closure of all fistulae. Fistulae remained closed 10 months later without any additional medication used in the interim. The third patient had three open fistula with two of three closed at week 6 and all closed at week 10 but developed a small abscess at week 11 and was discontinued. One patient had enterocutaneous fistulas secondary to CD and had a significant improvement in her mucosal disease with a marked decrease in her CDAI but without significant change in her fistulous output.


Thirteen of the initial 20 patients (65%) enrolled demonstrated a decrease in CDAI of at least 100 points which was maintained on at least two scheduled visits. Eleven patients (55%) continued to demonstrate a decrease of at least 70 points at week 12 and seven (35%) had a persistent decrease of at least 100 points. In addition, two patients who had responded increased their CDAI at this point and were not considered responders at week 12. Five patients (25%) were in remission at week 12. An additional patient maintained remission from weeks 4 to 10 but had an increase at week 12 as her dose was decreased as per protocol. The decrease in CDAI represented improvement in all categories including frequency of diarrhoea, abdominal pain, well-being and extra-intestinal manifestations (Table 2).

Response time varied with an initial improvement demonstrated within 1–3 weeks in most patients. Three patients showed a mild increase in symptoms (CDAI increasing 60 points or less) at weeks 1–2 before a marked benefit was observed. Among those completing the 12-week study, the average reduction in CDAI was 92 points (range: 36–249). As the dose was decreased due to an ANC above the target range, a modest increase in CDAI was seen in five responders although still well below the baseline in three. Response was lost in three patients as dose was decreased. Three of the four patients who were discontinued because of worsening disease had pancolonic disease, one of whom eventually went to surgery and was felt to have ulcerative colitis by surgical and pathological evaluation. However, other patients with colonic CD responded to rhuG-CSF treatment.

Subgroups of patients taking a particular medication (and may have been taking more than one) did not demonstrate any clear predictors of response or lack of response. Four patients were receiving no medication other than G-CSF. All responded with a decrease of >100 points with one achieving remission. Of the seven on prednisone, four achieved remission, three had no response. Of the 12 on sulfasalzine or mesalazine, five had no response, three a response but did not achieve remission and four went into remission. Whether benefits of G-CSF are blunted by concomitant therapy or synergistic with steroids or sulfasalzine or mesalazine cannot be determined by this study.

Medication adherence was >98% as assessed by return of empty vials. Adherence was confirmed by an elevation in ANC although the ANC does not allow a direct assessment of amount actually dosed. While the numbers are small and do not allow significant subgroup analysis, no trends could be identified with regard to response and concomitant medication, disease location, tobacco use or severity of disease.


This pilot study offers preliminary evidence that rhuG-CSF is a safe and potentially effective therapy for the treatment of active CD. About 25% (five of 20) of patients achieved remission during the study and additional patients demonstrated a significant improvement for an overall improvement observed in 55% (11 of 20). We observed a beneficial response in patients with fistulizing disease as well.

Adverse effects were limited and transient. A primary concern at the onset, this trial was the theoretical possibility that rhuG-CSF might exacerbate the disease by amplifying the inflammatory cascade. While three patients did discontinue because of worsening of disease, the relationship to the medication is not certain in two of the three with their discontinuation at week 2 and week 5 (patient 12 and patient 17). The third (patient 10) appeared to have a more direct worsening coinciding with initiation of the medication and discontinued at week 2. She failed additional therapy and eventually underwent a colectomy. While her previous history had been consistent with CD, pathological examination of the resected specimen suggested that this patient might not have had CD but ulcerative colitis. All three patients had primary colonic disease although other patients with colonic disease improved significantly in response to rhuG-CSF therapy. The patient hospitalized for a viral-like syndrome may have had drug-related toxicity although a similar syndrome has not been reported in the literature and no similar case is on file with the pharmaceutical company.

Interpretation of the results is complicated by the issue of dosing. The selection of the target ANC (25–35 × 109/L) in this study was guided by the choice of a range that would demonstrate a clear biological effect balanced with safety concerns. All but three patients underwent a dose reduction during the study. Coinciding with the dose reduction, some of the patients experienced a decrease in the therapeutic benefit as measured by the CDAI. Nevertheless, these patients CDAI still remained markedly below the baseline CDAI. This range of ANC was chosen as a surrogate marker for the desired augmentation of neutrophil function, although no functional neutrophil testing was conducted as part of this pilot study. A higher initial response in the ANC at week 1 was predicative of a beneficial response. The decreased response in some patients after the end of treatment and 4-week follow up (weeks 12–16) and diminished benefit for some in the final weeks of therapy may reflect that the benefit of G-CSF in many patients is short lived, stimulating a target cell, neutrophils, which have a short lifespan. However, the dose of most individuals was decreased early on and the duration and extent of benefit may have been limited by the dose-administered.

The optimal dose remains to be determined: if a higher dose would produce a greater response or whether, as with other biological therapy in CD, the dose–response demonstrates a bell-shaped curve, with higher doses being less effective. Because the medication was administered as a flat vial dose and was not on a weight-based dosing, several of the non-responders who were heavier, received a lower daily per kilogram dose. Overall, no significant difference in mcg/kg dose received was seen between responders and non-responders, although the number of patients was small. A weight-based dosing may be more appropriate. The duration of benefit is also a central concern with most patients experiencing an increase in disease activity within 2–4 weeks after discontinuing therapy. Maintenance therapy with rhuG-CSF was not explored in this study.

Few adverse events were noted. Most patients experienced some mild bone pain that was minimized by acetaminophen administered before injections. More significant bone pain and headaches were noted by several patients despite acetaminophen, although these reactions were transient. All bone pain resolved within a few weeks and with continued rhuG-CSF therapy during the treatment period. An extensive clinical experience exists for rhuG-CSF administered for other indications in over 2 million people. As this is often at higher doses than used in this study, unexpected significant adverse events are unlikely, although it has not been widely used in patients with presumably normal bone marrow, such as patients studied with CD.32 One serious adverse event in this trial, a viral-like syndrome, has not been reported with rhuG-CSF and its relation to drug is uncertain. The patient's initial increased ANC response to rhuG-CSF became blunted and had marked suppression of T4 T-cell subset without identification of any viral, bacterial or fungal agent. The patient's syndrome resolved fully within 2 weeks after hospitalization.

Endoscopic assessment was not performed in this study. Whether mucosal healing paralleled the clinical response is unknown. A recent report by Dejaco et al. suggested a benefit of G-CSF in mucosal healing. Five individuals with clinically inactive Crohn's but with active mucosal disease were treated in an open-labelled study with 300 mcg of rhG-CSF subcutaneously three times weekly for 12 weeks. Two of the five achieved complete mucosal healing (endoscopically assessed), one after 12 weeks of treatment, another 9 months after completion of therapy. An additional patient experienced closure of two fistulae as well.33 This report suggests that mucosal healing may be achieved in some patients.

This pilot study was undertaken to test a hypothesis that CD results from a primary functional deficiency of innate immunity. The approach developed through investigations of the association between CD and a variety of rare disorders, which involve qualitative and quantitative neutrophil dysfunction. Numerous case reports have detailed the coexistence of CD or a CD-like phenotype in patients with CGD, GSDIb, and autoimmune and cyclic neutropenia, as well as other innate immune defects.1–18 A serial mucosal immune response to bacterial invasion has been proposed in which serum factors and neutrophils remain the front line of defence. If these functions are unable to provide an adequate response, macrophages and, in turn, lymphocytes are activated. The activated Th-1 profile evident in CD may represent a secondary effort by the intestinal immune system to compensate for a more proximal insufficiency of innate immunity.19

Improvement in gastrointestinal symptoms has been noted in patients with these immunodeficiency diseases when treated with G-CSF and GM-CSF.8, 22–25 Crohn's associated with cyclic neutropenia was also reported to be maintained in remission by treatment with G-CSF.34 In addition, after the initiation of this study, a young Crohn's patient without an associated immunodeficiency was reported with significant improvement for fistulous disease during treatment with G-CSF. This patient presented at 8 years old with oral granulomatous lesions. He later developed perianal and enterocutaneous fistulae which improved significantly on a dose of 10 mcg/kg, a dose much higher than used in this pilot study.35

Inadequate neutrophil function has been previously characterized in CD including a deficiency of chemotaxis, response to complement, and microbicidal and candicidal killing.31, 36–41 These neutrophil abnormalities may represent the primary defect in CD which potentially stems from an immune deficiency rather than an excessively activated response. The 3020insC mutation in the NOD2 gene, recently identified to be linked with CD, is associated a deficient response of NFκ-B when stimulated by lipopolysaccharide.42 Unfortunately, CARD15 mutations, identified as associated with Crohn's after this trial, was not performed in these patients. This observation suggests a possible defect in the innate immune system may be responsible for at least some cases of CD. This model suggests a shift in the therapeutic paradigm from one of immunosuppression to focused immune stimulation for the treatment of CD directed towards improving innate immunity. Targeted stimulation of specific aspects of innate immune function may be most appropriate for a subgroup of patients having a defect in their innate immunity.

Both G-CSF and GM-CSF have been suggested to be beneficial in open-labelled pilot trials. Both rhuGM-CSF and rhuG-CSF were selected for study because of their known direct effects of its broad augmentation of neutrophil function, including chemotaxis, superoxide generation and bactericidal killing.27–29 While the benefit noted in this trial might be due to its influence on neutrophil function, as this therapy runs counter to most other therapies in CD, important questions include possible mechanisms of action. Functional and mechanistic studies were not performed in this trial. Unlike GM-CSF, G-CSF has been suggested to act through a number of other immunomodulatory pathways, possibly functioning not as a stimulant of innate immunity alone but also to suppress certain aspects of immune response.

More specifically, rhuG-CSF has been demonstrated to down-regulate interleukin (IL)-2, TNF-α and interferon (IFN)-γ.43 The rhuG-CSF has been suggested to influence T-lymphocyte cytokine production towards a Th-2 cytokine profile44 and to augment other anti-inflammatory cytokines such as IL-1RA and soluble TNF-receptors.45 It is unclear whether rhuG-CSF acts directly on effector cells or via indirect production of other cellular messengers to decrease proinflammatory cytokines. Receptors for rhuG-CSF have been identified on neutrophils, although their presence on other cell types is less well defined. G-CSF inhibited the release of TNF-α from monocytes when stimulated by lipopolysaccharide only when pretreated by G-CSF in the presence of neutrophils. TNF-α production by monocytes was not altered by G-CSF without the presence of neutrophils.43 The other cytokine modifying effects of rhuG-CSF may therefore occur through messengers generated in the neutrophil as a result of stimulation by rhuG-CSF directing the response of downstream events. Neutrophils may help to guide a coordinated mucosal response to potential microbial threats to the mucosa, underscoring a complex interaction between neutrophils, macrophages and lymphocytes that has not been fully characterized.

Whether rhuG-CSF acts in CD through augmenting innate immunity or by influencing cytokine production indirectly, the improvement of symptoms observed in this pilot study suggests that the role of the neutrophil in CD merits further investigation. The prevailing understanding of the role of the neutrophil in CD has been as a secondary factor, which contributes and amplifies excessive inflammatory activity. In the light of this trial and other supportive evidence, the role of the neutrophil in CD merits reconsideration. This initial pilot study provides data suggesting rhuG-CSF is a safe and possibly effective agent for the treatment of active mucosal CD as well as fistulous complications. Further study is required to elucidate these mechanisms and to validate the effect of rhuG-CSF for active CD in larger randomized-controlled trial.


Authors are gratefully acknowledge the contributions of Cathy Scott, RN and Arlyn Pittler, FNP in conducting this trial. Authors appreciate the insightful discussions of these ideas with Dr William Stenson. The study was supported by Amgen.

On the basis of the hypothesis and early clinical experience, Washington University applied for a patent, which is currently under review, covering the use of colony-stimulating factors for the treatment of Crohn's disease.