Approximately 10–20% of patients with acute Kawasaki disease (KD) do not respond to standard intravenous gammaglobulin (IVGG) therapy and have persistent or recrudescent fever after completion of the infusion.^1–,¹⁰ Additional IVGG therapy is commonly used for such patients,^1–,⁸ but a preventive effect on the development of coronary artery lesions (CAL) has not been proved by retrospective studies^1–,³ making a comparison between KD patients treated with additional IVGG and those not treated with additional IVGG. According to the study by Burns and colleagues,¹ among patients with persistent or recrudescent fever after initial IVGG therapy, CAL developed in five of 25 patients (20%) treated with additional IVGG and none of 21 patients (0%) not treated with additional IVGG. In the studies by Han and colleagues² and Durongpisitkul and colleagues,³ the prevalence of CAL tended to be higher in KD patients who received additional IVGG than in those who did not receive additional IVGG (43% v 27% and 21% v 9%, respectively). Other reports^4–,⁶ have also shown that the prevalence of CAL is high (33–54%) in patients treated with additional IVGG. This is probably because physicians are inclined to give additional IVGG to patients who appeared more ill and were believed to be at high risk for CAL.¹ Furthermore, we suspect that a delayed start of additional IVGG may fail to prevent CAL, since prolonged fever is a powerful predictor of CAL;¹¹ however, few previous studies^1–,⁶ have referred to the number of febrile days before start of the additional IVGG infusion. We therefore focused on KD patients treated with additional IVGG owing to unresponsiveness to initial IVGG, and studied the risk factors for CAL to test a hypothesis that additional IVGG therapy, if administered early, may decrease the prevalence of CAL.

METHODS

RESULTS

Of 816 patients with acute typical KD in our 12 hospitals in Japan, 746 received initial IVGG (1–2 g/kg) administration plus oral aspirin (30–50 mg/kg/day). Brands used for IVGG were freeze dried sulphonated product (Venilon), polyethylene glycol treated product (Venoglobulin IH or Glovenin I), and PH4 treated acidic product (Polyglobin N). A total of 81 (10.9%) patients were treated with additional therapies because of persistent fever (n = 65) or recrudescent fever (n = 16) after completion of the initial IVGG infusion. The fractions of the patients were not significantly different in any factors of initial IVGG infusion such as the number of illness days at the start, the dose, and the number of days that the dose was divided over (table 1⇓). For additional therapies, 48 patients were treated with IVGG (0.5–2 g/kg), 24 with IVGG combined with corticosteroids (1–2 mg/kg/day), ulinastatin (15 000 U/kg/day), or both, and nine with corticosteroids or ulinastatin. Furthermore, a third course of treatment was given to nine patients: IVGG (n = 1), IVGG with ulinastatin (n = 5), corticosteroids (n = 1), and corticosteroids with ulinastatin (n = 2). Overall, we gave additional IVGG without corticosteroids or ulinastatin to 46 patients, two of whom were excluded from the study because they received it more than 14 days after completion of the infusion of initial IVGG. Of 44 patients who met the inclusion criteria, 17 (38.6%) had CAL.

In univariate analyses, the significant risk factors for CAL (p < 0.05) were the number of febrile days pre-additional IVGG (⩾10 days), the number of febrile days post-additional IVGG (⩾2 days), the number of days that initial IVGG was divided over (4 or 5 days), the white blood cell count pre-additional IVGG (⩾16 000/μl), the C reactive protein concentration pre-additional IVGG (⩾9.7 mg/dl), and the white blood cell count post-additional IVGG (⩾11 300/μl) (table 2⇓). Data regarding the white blood cell count and the C reactive protein concentration post-additional IVGG were unavailable in five patients, all of whom had no CAL. The number of illness days at the start of initial IVGG infusion was same as the number of febrile days pre-initial IVGG in all patients. The number of illness days at the start of additional IVGG infusion (⩾10 days) was less significantly associated with CAL (p = 0.008) in comparison with the number of febrile days pre-additional IVGG (p = 0.002); hence, the number of febrile days pre-additional IVGG, rather than the number of illness days at the start of additional IVGG infusion, was subjected to multivariate analyses.

We performed a multiple logistic regression analysis for the variables significantly associated with the risk for CAL, excluding or including the white blood cell count post-additional IVGG separately because of unavailable data (table 3⇓). In both exclusion and inclusion of the white blood cell count post-additional IVGG, the only independent risk factor was the number of febrile days pre-additional IVGG (⩾10 days).

Of 17 patients with CAL, none was found to have CAL before the start of the initial IVGG infusion. CAL were detected in 14 patients (13 males and one female; age range 1–91 months) after start of the initial IVGG infusion before start of the additional IVGG infusion. Among the 14 patients, additional IVGG was given to three before 10 febrile days and to 11 on or after 10 febrile days. CAL were detected in three patients (a 2 month old female, an 8 month old male, and a 13 month old male), after the start of the additional IVGG infusion. Additional IVGG was given to all of the three patients on or after 10 febrile days.

No adverse events were observed in any patient treated with additional IVGG. There were no signs of hypotension, congestive heart failure, allergic reaction, or hyperviscosity.

DISCUSSION

In the present study, the number of febrile days pre-additional IVGG (⩾10 days) was an independent risk factor for CAL in patients with acute KD who did not respond to initial IVGG therapy and received additional IVGG therapy. It is possible that additional IVGG starting before the first 10 febrile days may prevent CAL. In this connection, let us recall the well known recommendation^7–,⁹ that initial IVGG should be given within the first 10 illness days. We believe that “febrile days”, rather than “illness days”, may be a better marker of the need for additional IVGG in recrudescent fever, on the basis of the result that the presence of CAL was more significantly related to the number of febrile days pre-additional IVGG than the illness days at the start of the infusion of additional IVGG. Few previous reports regarding additional IVGG therapy^1–,⁶ did not refer to the timing of administration of additional IVGG; the high reported prevalence of CAL may reflect a delayed start. Prolonged fever for 10 days or longer as a risk factor for CAL is supported by the following data. (1) In a pathological study,¹⁶ the formation of CAL began after 10 illness days. (2) In an echocardiographic study,¹⁷ CAL appeared at around 10 illness days. (3) In a clinical study,¹¹ patients with CAL had a significantly higher temperature than those without CAL during 10–13 illness days. Since additional IVGG appeared to be safe without apparent adverse effects,⁴ we suggest that it should be administered before the first 10 febrile days to initial IVGG non-responders.

Our results indicated that age, sex, white blood cell count, and C reactive protein concentration were not independent risk factors for CAL. On the other hand, age less than 6 months^14,15 and male sex¹⁵ were reported to be risk factors for CAL in KD patients treated with IVGG. It has also been reported that CAL were associated with abnormal laboratory data pre-initial IVGG, including high counts of neutrophils and bands,^3,10 and abnormal laboratory data post-initial IVGG, including increased white blood cell count and increased C reactive protein concentration.^18,19 This may be attributed to subject differences: the present study concerned only patients treated with additional IVGG, while previous studies^{3,10,14,15,18,19} concerned all KD patients treated with IVGG. There is a possibility that additional IVGG reduced the risk factors for CAL in initial IVGG non-responders, such as age <6 months, male sex, high white blood cell count, and high C reactive protein concentration in the present study.

In order to start the infusion of additional IVGG before the first 10 febrile days, we make three recommendations to physicians who treat patients with acute KD. First, KD patients should be treated immediately with an initial IVGG infusion. Although the efficacy of early treatment before five days of illness is controversial,²⁰ Tse and colleagues²¹ reported that treatment with initial IVGG on or before the fifth illness day resulted in better coronary outcome. Second, initial IVGG should be given in a single infusion, because 4–5 daily infusions led to delayed administration of additional IVGG. In Japan, the medical insurance system previously prevented physicians from treating with a single infusion despite the proven efficacy,⁹ but it has recently allowed its use. Third, it should be judged early whether patients respond to initial IVGG or not. However, we agree with Burns and colleagues¹ and Brogan and colleagues⁷ that patients with fever at least 48 hours after completion of a single infusion of initial IVGG are considered non-responders, since occasional responders have fever during the period. If additional IVGG is given 48 hours after completion of the infusion before the first 10 febrile days, initial IVGG should be completed by the first seven febrile days.

The present study has some limitations. There may be unknown selection bias owing to the retrospective nature of the study. A prospective randomised control study would be optimal to establish the efficacy of additional IVGG for patients who do not respond to initial IVGG, but it would be difficult to justify inclusion of an untreated control group in view of the increased risk of CAL in patients with persistent or recrudescent fever. We used four or five daily infusions of IVGG, which is not currently recommended in other countries.^7–,⁹ The administration of various brands of IVGG may be a confounding factor.²⁰

In our study, among KD patients with persistent or recrudescent fever after initial IVGG therapy, administration of additional IVGG before the first 10 febrile days was associated with a decreased risk for CAL, when compared with the risk in those who were retreated later. This finding provides support for early diagnosis and initial treatment of KD and suggests that an additional IVGG infusion, if administered early, may prevent CAL in initial IVGG non-responders.