In a child presenting with features consistent with a diagnosis of juvenile idiopathic arthritis, what clinical features or laboratory findings (at presentation) predict a diagnosis of acute lymphoblastic leukaemia?


  • Damien McKay,

    1. Department of Rheumatology, The Children's Hospital at Westmead, Sydney, New South Wales,
    2. Discipline of Paediatrics and Child Health, University of Sydney, Sydney,
    Search for more papers by this author
  • Louisa Adams,

    1. Department of Rheumatology, The Children's Hospital at Westmead, Sydney, New South Wales,
    Search for more papers by this author
  • Genevieve Ostring,

    1. Department of Rheumatology, The Children's Hospital at Westmead, Sydney, New South Wales,
    Search for more papers by this author
  • Davinder Singh-Grewal

    Corresponding author
    1. Department of Rheumatology, The Children's Hospital at Westmead, Sydney, New South Wales,
    2. Discipline of Paediatrics and Child Health, University of Sydney, Sydney,
    3. NSW Department of Immunology, Allergy, Infectious Diseases, Sydney Children's Hospital, Randwick,
    4. The John Hunter Chidren's Hospital, Newcastle and
    5. The University of New South Wales School of Women's and Children's Health, Sydney, New South Wales, Australia
    Search for more papers by this author

Dr Davinder Singh-Grewal, Department of Rheumatology, The Children's Hospital at Westmead, Locked Bag 4001, Westmead NSW, Australia. Fax: +618 9845 3432; email:

Clinical Scenario

A 10-year-old boy presented with joint pain and arthritis with joint effusions and decreased range of motion in the right wrist, both elbows, and right ankle. Initial investigations showed a normal full blood count (FBC) and film (haemoglobin 124 g/L, white cell count (WCC) 8.1 × 109/L, platelets 378 × 109/L) with raised inflammatory markers (ESR 78 mm/h, CRP 40 mg/L). Neither antinuclear antibodies nor human leukocyte antigen B-27 were present. X-rays demonstrated a right ankle effusion but were otherwise normal.

The child continued to have joint pain and swelling along with pain and inflammation at the insertions of the Achilles tendons at the ankle and plantar fascia over the metatarsal heads consistent with enthesitis. After 6 weeks of symptoms, he was diagnosed with enthesitis-related juvenile idiopathic arthritis (JIA), a form of the disease which predominantly affects boys with an onset of symptoms between 9 and 12 years of age.1

After an initial improvement with indomethacin therapy, his symptoms became progressively worse. He received 2 weeks of oral corticosteroids, and intra-articular corticosteroid injections to both wrists with good symptomatic relief.

Five months after initial presentation, the child's appetite deteriorated, and he experienced significant weight loss, became pale and fatigued. Repeat FBC demonstrated haemoglobin of 48 g/L, WCC of 2.3 × 109/L and platelet count of 28 × 109/L. A bone marrow aspirate was performed and was diagnostic of acute lymphoblastic leukaemia (ALL).

Structured Clinical Question

In a child presenting with features consistent with a diagnosis of JIA, what clinical features or laboratory findings (at presentation) predict a diagnosis of ALL?

Search Strategy

A search was performed on MEDLINE (from January 1966 to January 2009), CINAHL (from January 1982 to January 2009), EMBASE (January 1980 to January 2009), Cochrane Central Register of Controlled Trials (March 1996 to January 2009) and SportDiscus (January 1985 to January 2009). The search was undertaken using the following MeSH headings.

  • 1JIA or juvenile rheumatoid arthritis (JRA) or juvenile chronic arthritis or arthritis; and
  • 2leukaemia or leukaemia or malignancy

Results were limited to children. Abstracts and whole papers were read where required to select the most appropriate studies to answer the clinical question. Details of the studies identified by the literature search and considered most applicable are listed in Table 1.

Table 1.  Studies identified in literature search applicable to the clincial question
ReferenceStudy populationStudy design & level of evidence (GRADE criteria)ResultsComments
  1. ALL, acute lymphoblastic leukaemia; ESR, erythrocyte sedimentation rate; JIA, juvenile idiopathic arthritis; JRA, juvenile rheumatoid arthritis; LDH, lactate dehydrogenase; sJRA, systemic juvenile rheumatoid arthritis; WBC, white blood cells; WCC, white cell count.

Jones et al.271 children with ALL
206 controls with JRA
Retrospective case-control
Low evidence
• Strongest predictors for ALL were low WCC, low-normal platelet count and night-time pain
• Presence of multiple predictors increased sensitivity and specificity for diagnosis of ALL
• LDH raised in 77% of ALL versus 38% of JIA patients
• LDH values (>2× normal) in the ALL group
• Specifically considers ALL and JIA
• Unclear if patients with ALL initially fulfilled the diagnostic criteria for JIA
• Identifies distinct diagnostic predictors
• Multicentre study
• Large sample size, appropriate statistical analysis
Ostrov et al.310 children with acute leukaemia
10 children with sJRA (JIA)
Retrospective matched case-control study
Low evidence
• Bone and night pain common in leukaemic patients
• High spiking fever, morning stiffness and rash in 80–90% of sJRA and 10–20% of leukaemia patients
• Polyarthritis in 6 of 10 sJRA patients versus 1 of 10 leukaemic patients
• No differences in lab findings at presentation
• Compares groups at presentation
• Less applicable as considers specific subgroup of JIA
• Cases and controls' age and sex matched
• Small sample size, appropriate statistical analysis
Robazzi et al.4406 patients with acute leukaemiaCase series
Very low evidence
• Osteoarticular manifestations present in 54.5%
• Large joints more than small joints
• Single joint involvement more common
• WBC normal range in 52.5% of patients
• Platelet count <150/mm3 in 84%
• ESR raised in most cases
• Highlights the high frequency of acute leukaemia patients with musculoskeletal symptoms and potential for misdiagnosis
• Patients not initially diagnosed with arthritis
• No control group with JIA
• Large sample size, appropriate statistical analysis
Gonçalves et al.59 children with diagnosis of malignancy (leukaemia in 6)Case series
Very low evidence
• Arthritis most frequent presenting symptom in leukaemic patients
• No consistent examination findings
• Anaemia in 5 of 6 leukaemia patients
• Thrombocytopenia in three leukaemia patients
• LDH increased in 3 leukaemia patients
• Identifies the frequency of arthritis as a presenting symptom of leukaemia
• Small sample size
• No statistical analysis performed
Barbosa et al.661 patients with leukaemiaCase series
Very low evidence
• Frequency of symptoms: limb or joint pain (62%); daytime pain (50%); and night pain (10%)
• Arthritis on examination in 13% of patients
• LDH >400 IU in 93% of patients tested, anaemia in 88%, leukopenia in 65.6%, leukocytosis in 42.6%
• ESR raised in all tested (7/7)
• Highlights frequency of musculoskeletal manifestations in leukaemia
• It is not stated whether there was an initial diagnosis of JIA
• No statistical analysis performed
Trapani et al.710 patients with malignancy, (6 cases of leukaemia)Case series
Very low evidence
• Leukaemia patients: presenting symptoms were fever (6/6 patients), malaise (4/6), monoarthritis (3/6), arthralgia (3/6) and night sweats (2/6)
• Anaemia and raised ESR in 5/6
• LDH increased in 3/3 patients measured
• Examines the issue of occult cancer presenting to rheumatology service
• Small sample size
• No statistical analysis performed
Cabral and Tucker829 children with malignancyCase series
Very low evidence
• 6 of 11 patients with peripheral musculoskeletal symptoms had leukaemia
• 6 of 9 patients with prolonged fever had leukaemia
• 19 of 29 had more than one abnormal blood test
• Results not provided for specific malignancies or patients
• No detail of abnormality in lab investigations
• No statistical analysis performed

What Is the Best Paper to Answer My Question?

The case-control study by Jones et al.2 is the paper that most closely addresses the clinical question and represents the best available evidence from published studies to answer this ques- tion. The Jones et al.2 study provides low-level evidence using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) criteria9; a system for grading the quality of evidence and the strength of recommendations that can be applied across a wide range of interventions and contexts including observational studies.

The other studies identified in the literature search were either not as applicable to the clinical question or were small case series and provided a very low level of evidence using the GRADE system.

The study by Jones et al.2 compares the clinical and laboratory characteristics of a group of 71 children with an eventual diagnosis of ALL (cases) who initially presented to a paediatric rheumatology service with musculoskeletal symptoms to those of a group of 206 patients with JIA (controls) randomly selected from patient populations at seven different paediatric rheumatology centres. We will use the term JIA for these control patients, although the authors use the term JRA, which is a term derived from a comparable but superseded classification system. The case and control groups were not matched for any demographic characteristic. Age, gender and race for each of the two groups were presented. While there was no significant difference in age between the case and control groups, no further statistical comparison of demographics between the case and control groups was undertaken, making it difficult to assess whether bias is introduced through the selection of these groups.

Unfortunately, it is not possible to establish whether the cases fulfilled the diagnostic criteria for JIA at presentation. If cases presented with musculoskeletal symptoms other than arthritis, they would not fulfil the criteria for a diagnosis of JIA, meaning that the groups were drawn from different populations, because the controls all had an established diagnosis of JIA. This selection bias would reduce the validity of the study and, under these circumstances, the case group would be significantly different from the patient described in our clinical scenario, who presented with clinical features fulfilling the diagnostic criteria for JIA, and would make the study less relevant to the clinical question that is being posed.

Furthermore, the criteria upon which the diagnosis of JIA was made were not explicitly stated, and it is not clearly stated whether the diagnosis of ALL was based on bone marrow biopsy. Without this information, the level of evidence provided by this study is further eroded.

A retrospective chart review was performed using a questionnaire sent to clinicians to collate clinical and laboratory findings present during the initial visit for both cases and controls. The definition of the diagnostic markers examined to predict blast negative ALL was clear, as was the way in which they were ascertained. One strength of this study is the relative absence of missing data. However, information on plasma uric acid was not available in a significant proportion of cases and controls, thus reducing the validity of uric acid levels as a predictor of ALL.

There were large numbers of patients in both the case and control groups, but a power calculation was not provided to verify that the study was appropriately powered. A statistical analysis of the data was performed using two-sided Student's t-test, a two-sided Pearson χ2 or Fisher's exact test where appropriate to determine if a presenting sign, symptom or diagnostic test was different between the JIA and ALL groups. Significance level was adjusted using the Bonferroni correction for multiple comparisons.

How Would I Do the Research Better?

Conducting a similarly designed retrospective case-control study could help establish the predictors of ALL if care was taken to ensure the two groups differed only in their final outcome (i.e. the final diagnosis of ALL or JIA). Comparing patients presenting with symptoms fulfilling diagnostic criteria for JIA in both the case and control groups would provide a more robust study design and increase the validity of the results.

A prospective cohort study would provide a stronger level of evidence. Prospective studies have the benefit of avoiding the reliance upon the accuracy of written record or recall of individuals that a retrospective design does. However, this would be impractical and costly, given the low event rate of ALL after presentation with symptoms consistent with JIA.

How Would I Apply the Information to the Patient?

Of the 71 patients with ALL, 25% had blasts in the initial blood smear and 75% (as in the case presented) did not. The three strongest single predictors for blast negative ALL were:

  • 1low WCC count (85% sensitivity/83% specificity),
  • 2low-normal platelet count (82% sensitivity/87% specificity), and
  • 3night pain (49% sensitivity/90% specificity).

Thus, WCC or platelet parameters are relatively good predictors of malignancy in this situation. Night pain has a poor sensitivity, so the absence of this symptom is not useful in ruling out malignancy as there are many false negatives. However, the presence of night pain is a relatively strong predictor of a diagnosis of ALL as there are few false positives.

The authors found that the presence of any two low haematological values (i.e. low-normal platelet count, depressed WCC or low haemoglobin) had a higher sensitivity (96%) and specificity (88%). The combination of night pain and any two low haematologic values had a sensitivity of 100% and specificity of 85%.

The values of Lactate Dehydrogenase (LDH) were increased in 77% of ALL patients and in 38% of JIA patients. However, the sensitivity of an elevated LDH for diagnosis of ALL was low at 53%.

Our child had a 5-month delay between initial symptoms and final diagnosis. At the time of diagnosis with ALL, prominent constitutional symptoms were present and counts were markedly depressed in multiple cell lines. Investigations at presentation did not demonstrate either a low WCC or low-normal platelet count, nor was there a history of night pain. Thus, the predictive features identified by Jones et al.2 would not have aided in establishing a diagnosis of ALL at initial presentation in our patient. As an LDH was not performed in our child at presentation, it is not possible to comment on whether this may have aided in an early diagnosis of ALL.

Clinical Bottom Line

The lack of high-quality evidence in answering this clinical question is typical of many such questions in paediatric rheumatology and in general paediatrics. We have been able to access only low-level evidence from the published literature, although many aspects of the chosen study closely address the clinical question and provide some guidance for clinicians.

As JIA is a diagnosis of exclusion, malignancy should be considered in the differential diagnosis of any child presenting with arthritis or arthralgia. In the majority of patients with acute leukaemia presenting with musculoskeletal symptoms, blasts are not present in the peripheral smear, making early diagnosis more challenging. Attention to specific symptoms such as night pain, severe joint pain, pain resistant to analgesics or non-articular bone pain is essential. Findings on musculoskeletal examination may be of little benefit in differentiating between the two diagnoses. In addition to an FBC and peripheral film, inflammatory markers, LDH and imaging of the skeletal system should be considered in the initial work-up of any child being diagnosed with JIA. The presence of any degree of anaemia, leukopenia or thrombocytopenia (even when single cell lines are involved) warrants further consideration. Significantly increased LDH should also alert the clinician to the possibility of a more sinister diagnosis. Lastly, any patient who does not respond to therapy as expected or develops new symptoms of concern should always have their diagnosis reconsidered.