Introduction

Stem cell transplantation (SCT) is a widely used treatment for hematological disorders, malignant solid tumors, and primary immunodeficiencies. Complications occurring within 100 days of transplantation can be a major cause of transplant-related mortality. Therefore, the prediction and prevention of such complications are extremely important.

Diffuse hyperpigmentation commonly occurs after chemotherapy and SCT. Hyperpigmentation usually appears after 1-4 weeks of chemotherapy, especially with alkylating agents, such as cyclophosphamide, ifosfamide, busulfan, and thiotepa. It may be generalized or localized at sites, such as the nails and teeth^{1, 2}. A previous study reported cutaneous symptoms, such as diffuse erythema, desquamation, and hyperpigmentation, in approximately 80% of children who underwent autologous SCT with thiotepa³. Because a variety of SCT complications manifest as early signs of skin reactions to chemotherapy, radiation, graft-versus-host disease (GVHD), and infection, skin monitoring is considered very important⁴. However, there have been very few reports on the relation between skin reactions and treatment-related complications, and the cause of pigmentation after chemotherapy or SCT. Skin pigmentation is thought to result from melanin accumulation in the dermal layer, caused by increased melanin synthesis and the destruction of the basement membrane due to inflammation. Melanin pigments are classified into two types: brown-to-black eumelanin and yellow-to-reddish-brown pheomelanin. The ratio of the two types of melanin, brown-to-black eumelanin and yellow-to-reddish-brown pheomelanin, determines the skin and hair color. The metabolic pathways involved in the production of these two melanin types are shown in Figure 1. After the oxidation of tyrosine by tyrosinase to dopaquinone, eumelanin is produced by the oxidation of 5,6-dihydroxyindoles as an intermediate, while pheomelanin is formed by the oxidation of 5-S-cysteinyldopa (5SCD) as an intermediate⁵. Some 5SCD leaks into the blood and is excreted in urine after being metabolized in the liver. Although urinary 5SCD levels depend on age, serum 5SCD values are independent⁶. Serum 5SCD levels have been reported to be related to hyperpigmentation in patients undergoing hemodialysis^{7, 8} and are widely used as biomarkers for malignant melanoma^9–11.

As 5SCD has not been reported as an SCT-related biomarker so far, we serially examined serum 5SCD levels before and after SCT to determine its association with SCT-related complications.

Patients and Methods

We prospectively analyzed 41 patients (27 males and 14 females) with a median age of 7.9 years who underwent SCT between May 2011 and March 2015 at Hokkaido University Hospital and provided informed consent or parental permission for enrollment in this study. This study was approved by the Institutional Review Board and Ethics Committee of Hokkaido University Hospital (Approval Number 012-0376). Serum samples were obtained from the patients using a non-invasive backflow method from a central venous catheter before conditioning therapy, on day 0 (the day of stem cell infusion), and on days 5, 10, 15, 25, and 40. Serum samples were obtained at these seven time points to monitor changes associated with conditioning regimens, engraftment, and complications, such as acute GVHD, based on the values before conditioning therapy. All blood samples were centrifuged at 3,000 rpm for 15 min and stored at －80°C until use.

We analyzed the association of eight 5SCD values (at each of the six time points, the highest value, and the difference between the baseline level of 5SCD before conditioning therapy and the highest level after SCT) with the following: conditioning regimen, donor source, malignant disease (having received some cycles of chemotherapy before SCT), and the onset of SCT-related complications within 100 days after SCT, such as acute GVHD, acute cutaneous GVHD, viral reactivation, pre-engraft syndrome (PES), and death not attributed to the primary disease. We used multivariate analysis to examine the independent association between serum 5SCD levels with the items associated with SCT-related complications in the univariate analysis.

Statistical analysis

Univariate logistic regression was used to evaluate the association between each 5SCD and SCT-related complications occurring within 100 days post-SCT, conditioning regimen, donor source, and HLA disparity. These factors were further assessed using multivariate logistic regression models to analyze the relation between the 5SCD concentration and SCT-related complications. Study confounders were selected based on a literature review and change-in-estimate criteria, which were set at a value greater than 10%. Potential confounding variables considered in the analysis included age, stem cell characteristics, HLA disparity, anti-thymocyte globulin use, steroid use, and conditioning intensity. P was set at p < 0.05. All statistical analyses were performed using JMP, version 14 (SAS Institute Inc., Cary, NC, USA).

Results

Patient characteristics

The median age of the patients at transplantation was 7.9 years (range: 0-22 years). None of the patients had renal failure or were undergoing hemodialysis. The patient characteristics are shown in Table 1. Indications for SCT included malignant diseases in 32 patients (hematological malignancies in 21 patients and malignant solid tumors in 11 patients) and non-malignant diseases in nine patients. Thirty-four patients underwent allogeneic SCT, and seven patients received autologous SCT. Myeloablative conditioning was provided to 30 patients and reduced-intensity conditioning was provided to 11 patients. The conditioning regimen, defined as myeloablative conditioning, included intravenous busulfan > 7.2 mg/kg, melphalan (> 140 mg/m²), and total body irradiation (> 8 Gy). The conditioning regimens included total-body irradiation (n = 16), busulfan (n = 14), cyclophosphamide (n = 16), and melphalan (n = 22). Twenty patients were administered steroids for GVHD or PES.

SCT-related complications

PES was observed in 18 patients. Acute GVHD was observed in 19 patients, of whom 16 had acute cutaneous GVHD and 12 had grade II-IV acute GVHD. Viral reactivation was observed in 21 patients, including that of cytomegalovirus (n = 13), Epstein-Barr virus (n = 4), varicella-zoster virus (n = 4), adenovirus (n = 2), and human herpes virus-6 (n = 1) (Table 2). Some patients showed reactivation of multiple viruses, but none showed reactivation of human herpes virus-6 or varicella-zoster virus alone. Thirteen patients relapsed or died, two of whom died of SCT-related complications within 100 days post-SCT. One patient with a solid tumor developed hepatic veno-occlusive disease and adenoviral infection after autologous SCT and died 60 days after SCT. Another patient with acute lymphoblastic leukemia had severe acute GVHD with adenoviral infection and died 87 days post-SCT.

Serum 5S-cysteinyldopa level

The average value of 5SCD reached two peaks, one on day 0 and the other on day 5, regardless of the stem cell source and conditioning intensity (Figure 2). Univariate analysis revealed that the 5SCD level on day 0 was associated with viral reactivation (p = 0.044) (Figure 3A), the 5SCD level on day 5 was associated with PES (p = 0.041) (Figure 3B), and the highest 5SCD level was associated with malignant disease (p = 0.038). In patients receiving allogeneic SCT (n = 34), both the peak level of 5SCD and the difference between the highest 5SCD and 5SCD level before SCT were associated with acute GVHD (p = 0.03 and 0.016, respectively) (Figure 4A and B). Both the 5SCD level on day 5 and the difference in 5SCD levels were associated with acute cutaneous GVHD (p = 0.030 and 0.026, respectively) (Figure 4C and D). There was no significant association between any serum 5SCD level and chronic GVHD, severity of acute GVHD, intensity and content of conditioning regimens, graft source, immunosuppressive agents, or death not caused by the primary disease.

Subsequently, we conducted multivariate analysis to examine the independent association between viral reactivation, acute GVHD, acute cutaneous GVHD, and serum 5SCD levels (Table 2). The 5SCD level on day 0 was independently associated with viral reactivation (odds ratio [OR], 3.32; 95% confidence interval [CI] 1.07-10.21, p = 0.002) (Figure 3A). Furthermore, the 5SCD level on day 5 and PES were associated independently (OR: 2.18; 95% CI 1.11-4.26, p = 0.007) (Figure 3B).

We found that both the highest level of 5SCD and the difference between the highest level of 5SCD after SCT and the baseline level of 5SCD before SCT were independent biomarkers for acute GVHD (OR: 2.01; 95% CI 1.06-3.79, p = 0.014 and OR: 1.90; 95% CI 1.04-3.45, p = 0.015, respectively) (Figure 4A and B). The 5SCD level on day 5 and difference of 5SCD levels were independent biomarkers for acute cutaneous GVHD (OR: 2.71; 95% CI 1.09-7.53, p = 0.011 and OR: 2.34; 95% CI 1.11-4.52, p = 0.005, respectively) (Figure 4C and D).

Discussion

SCT is used to treat and cure many diseases, including childhood diseases. Predicting and controlling SCT-related complications are extremely important to prevent transplant-related mortality. We found that serum 5SCD can serve as a biomarker for SCT-related immunological complications, such as acute GVHD, viral reactivation, and PES. Since diffuse hyperpigmentation is commonly observed in patients receiving chemotherapy or SCT, we investigated whether the serum level of 5SCD, a precursor of pheomelanin, a melanin pigment, could be a biomarker for SCT-related complications. Our results indicated no association between the 5SCD levels and conditioning regimens. Therefore, we assumed that the elevation of serum 5SCD levels was induced by immunological damage rather than by cytotoxic damage during conditioning therapy and presumed that the skin was damaged, resulting in the production of pheomelanin.

GVHD and PES are immune-mediated reactions and major complications after allogeneic SCT, respectively. Acute cutaneous GVHD is characterized by erythematous macular and papular rashes¹², while PES is associated with skin rashes and capillary leak syndrome during the early engraftment phase¹³. Both GVHD and PES are accompanied by immunological skin inflammation. We speculate that this inflammation promotes pheomelanin production, resulting in elevated serum 5SCD levels. Further pathological examination of melanin pigments in patients with GVHD and PES is required to confirm our hypothesis.

Although the risk factors for viral reactivation include the use of anti-thymocyte globulin, steroids, and cord blood transplantation¹⁴, we did not find an association between serum 5SCD and these risk factors. Viral reactivation was associated with the levels of serum 5SCD on day 0 prior to disease onset. This may be related to early events preceding viral activation. However, the association between varicella-zoster virus reactivation and serum 5SCD remains unclear, as varicella-zoster virus reactivation typically occurs three months after SCT¹⁵. We only measured serum 5SCD levels up to 40 days post-SCT. Additionally, although we demonstrated no association between the severity of acute GVHD and serum 5SCD levels, we could not determine the association between the severity of viral reactivation and serum 5SCD levels because of the small number of patients with severe viral reactivation.

Various biomarkers have been reported to indicate disease severity and serve as prognostic factors for SCT-related complications. Soluble ST2 (also known as interleukin-33 receptor), interleukin-2 receptor α-chain, and tumor necrosis factor receptor-1 have been identified as prognostic markers for acute GVHD, and cutaneous elafin expression has been identified as a prognostic marker for acute cutaneous GVHD^16–20. MxA levels correlate with viral infections²¹. Shah et al. reported that elevated procalcitonin levels may be a biomarker of PES²². Compared with these biomarkers, 5SCD is widely used in several clinical settings, and its serum levels may be useful in predicting multiple SCT-related complications.

The current study has some limitations. First, it is difficult to clearly demonstrate the relation between the extent of pigmentation and SCT-related complications. Murakami et al. examined the degree of hyperpigmentation in patients undergoing hemodialysis by measuring melanin and erythema indices, revealing a significant correlation with serum 5SCD levels, suggesting the accumulation of pheomelanin in the skin⁸. Further studies to determine whether the degree of pigmentation can be an indicator of SCT-related complications would be useful for preventing transplant-related mortality. Secondly, the sample size was small. Therefore, further investigations with larger sample sizes, including adults, could validate our findings. Finally, it was difficult to identify the activation site in the melanin synthesis pathway leading to an elevation in 5SCD levels, as the precursor of eumelanin could not be measured in our study.

In conclusion, serum 5SCD levels may be predictive biomarkers for SCT-related complications, such as acute GVHD and viral reactivation. Pheomelanin production is presumed to be induced by inflammatory or immunological processes during SCT. However, further prospective studies are required to confirm these findings.

Acknowledgments

We would like to show our greatest appreciation to Professor Emeritus Tadashi Ariga, who gave valuable comments and suggestions to our study. We also thank Jimei Zhao for assisting submission of the manuscript.

Author Contributions

YT and AI designed the study; HG and IY performed statistical analysis; YT, AI, MS, YC, and AM analyzed and interpreted the data; YT and AM wrote the manuscript.

Conflicts of Interest

IY received research fund from Nihon Medi-Physics and honoraria for lectures from Chugai Pharmaceucial Co., AstraZeneca, and Pfizer. AM received honoraria for lectures from Novartis, Ohara, and Servier.