Early diagnosis of sinusoidal obstruction syndrome after hematopoietic stem cell transplantation, with modified diagnostic criteria including refractory thrombocytopenia

Abstract Sinusoidal obstruction syndrome (SOS) is a fatal complication of hematopoietic stem cell transplantation (HSCT). Early diagnosis for SOS can improve clinical outcomes significantly. Here, we performed a retrospective study to investigate the Cairo diagnostic criteria, in which SOS was defined as the development of two or more in seven events, including transfusion‐refractory thrombocytopenia. Among 154 cases of allogeneic HSCT, 10 cases of SOS using the European Society for Blood and Marrow Transplantation criteria (EBMT16) as the reference standard were identified. The original Cairo criteria could diagnose SOS 5 days earlier than any other established criteria, with some false‐positive results (sensitivity = 100.0%; specificity = 72.2%). When the cutoff was set to three events for the Cairo criteria, the diagnosis of SOS could be made 3 days earlier than that using the EBMT16 criteria, with comparable precision (specificity = 86.1%). The accuracy of the Cairo criteria improved further when the cutoff point was set to four (specificity = 93.8%). The fulfillment of the Cairo criteria was associated with high mortality. Based on our results, the Cairo criteria were also considered clinically useful, especially at three or four cutoff points. Further studies are required to validate and refine the criteria.


INTRODUCTION
Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is widely administered as a curative treatment for various hematological diseases. However, allo-HSCT can cause potentially fatal complications, such as graft-versus-host disease and sinusoidal obstruction syndrome (SOS), previously known as a veno-occlusive disease. The incidence of SOS has been estimated at approximately 10%-15% after allo-HSCT [1][2][3]. Severe SOS with multiple organ failure (MOF) is associated with an extremely high mortality rate (> 80%) [3]. Some studies have demonstrated that early initiation of treatment for SOS is related to significant improvement in clinical outcomes [4][5][6]. Therefore, early diagnosis is necessary to treat SOS effectively.
Owing to the difficulty in definitive diagnoses by invasive procedures, such as liver biopsy immediately after HSCT, SOS had been diagnosed with the modified Seattle criteria and/or the Baltimore criteria as surrogate gold standards for a long time [7,8]. The Baltimore criteria contain the mandatory event of hyperbilirubinemia (≥ 2 mg/dL) for diagnosis of SOS. As the elevation of bilirubin can be a delayed manifestation in SOS [9], this event potentially leads to a delay in the diagnosis. In 2016, the European Society for Blood and Marrow Transplantation (EBMT) advocated the revised diagnostic criteria for SOS in adults (EBMT16 criteria) [9]. Using the EBMT16 criteria, "classical SOS" is diagnosed with almost identical events to the Baltimore criteria, including hyperbilirubinemia. Therefore, these criteria could also lead to a delay in diagnosis.
Recently, Cairo et al. have proposed their own set of original diagnostic criteria [10]. They defined SOS as the development of two or more of the seven events ( Table 1). The unique characteristics of their criteria are as follows: refractory thrombocytopenia and Doppler ultrasound findings. Furthermore, hyperbilirubinemia is not a mandatory event, and the threshold for hyperbilirubinemia is lower than those in other established criteria. While the "Cairo criteria" are expected to allow for earlier diagnosis compared to the other established criteria, no verification and validation studies have been performed yet. In this study, we verified the "Cairo criteria" by performing a retrospective investigation of patients who underwent allo-HSCT.

Definitions of SOS and related terms
The modified Seattle criteria define SOS as the development of two or more of the following events within 20 days after HSCT: (1) hyperbilirubinemia (> 2 mg/dL), (2) hepatomegaly and/or right upper quadrant pain, and (3) weight gain (> 2% compared to baseline weight pre-HSCT) [7]. The Baltimore criteria define SOS as the development of hyperbilirubinemia (≥ 2 mg/dL) and two or more of the following events within 21 days after HSCT: (1) hepatomegaly, (2) ascites, and (3) weight gain (≥ 5% compared to baseline weight pre-HSCT) [8]. The EBMT16 criteria define "classical SOS" as the development of hyperbilirubinemia (≥ 2 mg/dL) and two or more of the following events within 21 days after HSCT: (1) painful hepatomegaly, (2) ascites, and (3) weight TA B L E 1 Diagnostic criteria for SOS.

Established criteria Modified diagnostic criteria
Modified Seattle criteria [7] Baltimore criteria [8] EBMT16 criteria (classical SOS) [9] Cairo criteria [10] The presence of two or more of the following within 20 days The presence of hyperbilirubinemia: ≥ 2 mg/dL and two or more of the following within 21 days gain (> 5% compared to baseline weight pre-HSCT) [9]. We defined "classical SOS" using the EBMT16 criteria as the reference standard.
We configured SOS diagnosis by the "Cairo criteria" as the development of two or more of the following events within 21 days after HSCT: (1) hyperbilirubinemia beyond the upper institutional limits (1.5 mg/dL in our institution), (2) weight gain (≥ 5% compared to baseline weight pre-HSCT), (3) platelet transfusions consistent with refractory thrombocytopenia (24 h corrected count increment falling below 4.5 × 10 9 /L) [11], (4) hepatomegaly, (5) right upper quadrant pain, (6) ascites, and (7) reversal of portal venous flow (hepatofugal flow) detected on Doppler ultrasound of the liver [10]. Either of the following events would also be defined as SOS: a liver biopsy consistent with SOS or elevated portal venous wedge pressure by direct measurements.
MOF was defined as coexistence with renal failure (serum creatinine ≥ 3 × the upper limit of normal or dependence of dialysis) and/or respiratory failure (oxygen saturation < 90% on room air, requirement for oxygen supplementation, or ventilator dependence) [1]. Remission of SOS was defined as the resolution of all signs and symptoms of established SOS diagnostic criteria [1]. The severity of SOS for cases who fulfilled any of the established criteria was graded using the EBMT16 grading for adults [9]. Conditioning regimens were classified as myeloablative conditioning if any of the following regimens were used: total body irradiation > 8 Gy, intravenous busulfan ≥ 7.2 mg/kg, or melphalan ≥ 140 mg/m 2 . Other conditioning regimens were classified as reduced-intensity conditioning [12].

Statistical analysis
Categorical and continuous variables of case characteristics were compared using Fisher's exact test and the Mann-Whitney U test, respectively. The intervals of diagnosed days between each criterion and the EBMT16 criteria were compared using the Wilcoxon signedrank sum test. To compare the specificity of each criterion, the negative proportion of each test in non-SOS cases was compared using the McNemar test with Yates's continuity correction [13]. The incidence of the SOS estimated by each criterion was calculated using the method of cumulative incidence, considering non-SOS mortality as the competing risk. Using the receiver operating characteristic (ROC) curve, the optimal cutoff was determined as the point closest to the upper left corner.
Overall survival was estimated using the Kaplan-Meier method with the log-rank test. Statistical significance was defined as a two-tailed p value < 0.05. All statistical analyses were performed using R version 4.1.2 and EZR version 1.55 [14].

Case characteristics
Data from 135 patients (154 transplant cases) were analyzed. The median follow-up period for survivors was 1008 days (33-3823) after allo-HSCT. The characteristics of all the cases are shown in Table 2. The number of cases who developed "classical SOS" according to the EBMT16 criteria was 10 (6.5%). Grade 3 and 4 SOS was observed in two and eight cases, respectively, and no grade 1 or 2 SOS was observed. In our cohort, haploidentical-related transplantation was not performed.
Ursodeoxycholic acid for SOS prophylaxis was routinely used in our hospital during the physicians estimated it to be required. A summary of the treatment and clinical outcomes for the cases with classical SOS is shown in Table S1.

Evaluation of established diagnostic criteria and the Cairo diagnostic criteria
The median numbers of days on which SOS was diagnosed using each criterion are shown in Table S2. Among the 10 cases with "classical SOS," three were diagnosed earlier with the modified Seattle criteria than with the EBMT16 criteria ( Figure 1). Differences in the time of diagnosis between the modified Seattle criteria and the EBMT16 criteria were not statistically significant (median, 0.0 days; range, 0-9 days; p = 0.174). No differences were observed in the diagnosis time between the Baltimore and EBMT16 criteria. Among the 10 cases with classical SOS, nine were diagnosed significantly earlier with the original Cairo criteria than with the EBMT16 criteria (median, 5.0 days; range, 0-16 days; p = 0.008). Furthermore, seven cases were diagnosed TA B L E 2 Case characteristics.  The cumulative incidence of SOS diagnosed according to each criterion is shown in Figure 2. Non-SOS death as a competing risk was not observed within 21 days after allo-HSCT.
Considering "classical SOS" in the EBMT16 criteria as the reference standard of SOS, the sensitivity of the modified Seattle, Baltimore, and original Cairo criteria was 90.0%, 100.0%, and 100.0%, respectively ( Figure 3 and Table S3). The specificity of the modified Seattle, Baltimore, and original Cairo criteria was 87.5%, 99.3%, and 72.2%, respectively. The specificity of the original Cairo criteria was significantly lower than that of the modified Seattle criteria (p < 0.001).

Evaluation of the Cairo diagnostic criteria with various cutoff levels
To improve the performance of the Cairo criteria, we next examined these criteria when the cutoff point for the number of events was raised to various levels. When the cutoff point was set to three events, six cases were diagnosed significantly earlier by the Cairo criteria with three events than by the EBMT16 criteria (median, 3.0 days; range, 0-16 days; p = 0.036) (Figure 1). While four cases with classical SOS could also be diagnosed earlier by the Cairo criteria with three events than by the modified Seattle criteria, significance was not observed (median 0.0 days; range, 0-5 days; p = 0.089). When the cutoff point was set to four events, five cases could be diagnosed earlier by the Cairo criteria with four events than by the EBMT16 criteria, whereas significance was not observed (median, 0.5 days; range 0-9 days; p = 0.058).
The sensitivity of the Cairo criteria with cutoff three, four, and five events was 100.0%, 100.0%, and 70.0%, respectively ( Figure 3 and Table S3). The specificity of the Cairo criteria with cutoff three, four, and five events was 86.1%, 93.8%, and 100.0%, respectively. There was no significant difference in specificity between the Cairo criteria with three events and the modified Seattle criteria (p = 0.789). The specificity of the Cairo criteria with four events was significantly higher than that of the modified Seattle criteria (p = 0.027). The ROC curve for the Cairo criteria is shown in Figure 3. Regarding diagnostic precision, the optimal cutoff was identified as four. The diagnostic precision and diagnosed time estimated using each of the criteria are summarized in Table 3.
The proportions of the development of events in the Cairo criteria at the time of fulfilling each criterion are shown in Table 4. In this study, transfusion-refractory thrombocytopenia was frequently

DISCUSSION
The original Cairo criteria, defined as the development of two or more events, could diagnose SOS significantly earlier than any other established criteria. As it can lead to the early initiation of SOS treatment, these criteria are considered clinically valuable. The modified Seattle criteria could not diagnose SOS significantly earlier than the EBMT16 criteria could. Defibrotide, known as an effective medication for SOS, has resulted in significant improvement when treating patients with severe SOS [15]. Some studies have demonstrated that a delay in the initiation of defibrotide treatment of over 2 days could significantly decrease the proportion of survival after HSCT [4][5][6]. Therefore, early diagnosis is necessary to treat SOS effectively. The original Cairo criteria were superior to the established criteria in terms of the early diagnosis of SOS. However, the specificity of the original Cairo criteria was significantly lower than that of the modified Seattle criteria.
Pediatric diagnostic criteria for SOS advocated by the EBMT16, which resemble the Cairo criteria, showed similar results [16][17][18]. While a diagnosis with the pediatric criteria was on average 3 days earlier than with the Seattle criteria, the number of patients fulfilling these criteria was extremely high (44.8% of patients undergoing allo-HSCT) [17].
Similarly, diagnosis with the original Cairo criteria contained a high rate of false positives. As defibrotide may cause fatal side effects, such as hemorrhage [19], the precise diagnosis of SOS is also important.
Therefore, we next examined the Cairo criteria when the cutoff point was raised to various levels.
When the cutoff point was set to three events instead of two, the Cairo criteria could also diagnose SOS significantly earlier than the EBMT16 criteria. Diagnostic precision was comparable to that of the modified Seattle criteria. The interval of diagnosis between these criteria and the EBMT16 criteria was approximately 3 days. According to the study on defibrotide mentioned above, a 2-day delay in SOS diagnosis is associated with worse clinical outcomes [4][5][6]. not achieved, in terms of diagnostic precision, this cutoff was also considered clinically useful. When the cutoff point was set to five events or more, these criteria were considered clinically inadequate because of the deterioration in sensitivity. Taken together, the Cairo criteria were considered clinically useful, especially by using three or four event cutoff points. We propose that the development of three events in the Cairo criteria is considered "possible SOS," and the development of four events is considered "confirmed SOS." The term "confirmed" derives from the best diagnostic precision in the Cairo criteria. The term "possible" derives from the next best diagnostic precision and the possibility of early diagnosis. The use of these criteria might be associated with early and adequate treatment of SOS and improvements in clinical outcomes.  [10]. For early diagnosis, the inclusion of this event may require some refinement. Other ultrasound approaches, such as HokUS-10, seem to be helpful [22,23].
Clinical literature has suggested that SOS without hyperbilirubinemia is uncommon in adults, unlike in children [7,16] Our study has some limitations. First, because it was a small retrospective trial conducted in a single institution, this study lacked sufficient power. Accordingly, the benefits of the Cairo criteria using a cutoff of three or four events compared with the modified Seattle criteria need further investigation and confirmation. Second, none of the cases with SOS were definitively diagnosed by liver biopsy or autopsy. According to a previous study on patients with SOS undergoing a hepatic biopsy, the sensitivity and specificity of the Baltimore criteria were estimated to be 56% and 92%, respectively [21]. In our study, the EBMT16 criteria were chosen as surrogate reference standards. Therefore, the true diagnostic precision of the Cairo criteria remains unclear. Third, late-onset SOS was not investigated in this study. The original Cairo criteria do not limit the SOS onset date. In our study, we followed the EBMT16 criteria in which "classical SOS" was defined within 21 days after HSCT. Further studies are needed to evaluate these criteria in late-onset SOS. Despite these limitations, our study is valuable as the first verification of these criteria.
In conclusion, the Cairo diagnostic criteria could diagnose SOS earlier than the established criteria, with comparable precision. These criteria could be useful for the early and precise diagnosis of SOS, potentially leading to early initiation of treatment and improvement of clinical outcomes. We propose that the development of three events in the present criteria is considered "possible SOS," and that the development of four events is considered "confirmed SOS."

AUTHOR CONTRIBUTIONS
HI