Prediction of Severe Cisplatin-Induced Neutropenia Using Serum Albumin Concentration: A Retrospective Study

Background: Hypoalbuminemia may occur in patients with gastric cancer because of reduced food intake and gastric dysfunction. Cisplatin that is used in the treatment of gastric cancer not only has gastrointestinal side effects but also has a high serum protein-bound fraction, which causes alteration in pharmacokinetics, especially reduction of serum albumin concentration that may increase the risk of cisplatin-induced neutropenia. Hence, alteration of serum albumin concentration poses a major safety issue during anticancer therapy. We examined whether uctuations in serum albumin concentration is a risk factor for predicting the development of severe neutropenia. Methods: This retrospective study included patients with gastric cancer undergoing cisplatin plus S-1 combination therapy to analyze their serum albumin concentration and the frequency of grade 3–4 neutropenia. We then investigated the relationship between the serum albumin concentration before cisplatin administration in the treatment course during which the neutrophil count reached nadir and the neutrophil count uctuation after cisplatin administration. Results: Grades 3–4 and 0–2 neutropenia developed in 24 and 48 patients, with a mean serum albumin concentration of 3.60 ± 0.54 and 3.77 ± 0.51 g/dL, respectively, before chemotherapy (P = 0.23). During chemotherapy when the neutrophil count reached nadir, the serum albumin concentration levels before cisplatin administration were 3.39 ± 0.60 and 3.85 ± 0.59 g/dL, respectively; the grade 3–4 neutropenia group had a signicantly lower concentration of serum albumin than the grade 0–2 neutropenia group (P = 0.006). Lower serum albumin concentrations before cisplatin administration were signicantly correlated with a decrease in neutrophil count after cisplatin administration (r = 0.463, P < 0.001). According to the receiver operating characteristic curve analysis, patients with serum albumin concentrations below 3.25 g/dL before cisplatin administration exhibited a signicantly higher incidence of grade 3–4 neutropenia (odds ratio: 4.33).


Background
Serum albumin is a typical marker of nutritional status. A decrease in albumin synthesis caused by reduced food intake and gastric dysfunction can lead to hypoalbuminemia. Patients with gastric cancer may be susceptible to hypoalbuminemia, not only because of organic disease but also because of gastrointestinal side effect induced by anticancer drugs. The 2018 Japanese gastric cancer guidelines (version 5) recommends cisplatin plus S-1 combination therapy as the rst-line treatment for advanced/recurrent or unresectable gastric cancer; the recommendation is based on the trial results of SPIRITS [1] and JCOG 9912 [2]. Cisplatin is a therapeutic anticancer drug with a high emetic risk, commonly causing poor appetite after administration. It also exhibits a high serum protein-bound Page 3/13 fraction (> 90%) [3,4]. Therefore, patients with decreased serum albumin concentration receiving cisplatin are highly prone to cisplatin-induced side effects, particularly hematological toxicity caused by altered pharmacokinetics. Among hematological toxicities, severe neutropenia is a serious condition that has been frequently observed and it may progress to febrile neutropenia. Altered serum albumin concentration can be a major safety concern during anticancer therapy. However, this concentration is not currently recognized as a reference for optimizing chemotherapy. Clarifying this potential association may facilitate appropriate chemotherapy; hence, this study aimed to examine whether uctuation in serum albumin concentrations is a risk factor for predicting the development of severe neutropenia induced by cisplatin plus S-1 therapy.

Study design
Conducted at Jikei University Hospital in Tokyo, Japan, this retrospective observational study investigated the association between serum albumin concentration and the frequency of grade 3-4 neutropenia in patients with gastric cancer undergoing cisplatin plus S-1 combination therapy.

Patients and methods
This study examined the data of patients with advanced/recurrent or unresectable gastric cancer who received cisplatin plus S-1 combination therapy between March 2012 and March 2019. In accordance with the 2018 Japanese gastric cancer guidelines (version 5), these patients received 60 mg/m 2 concentration of cisplatin intravenously on day 8 plus 80 mg/m 2 concentration of S-1 orally on days 1-14 in a 21-day cycle every 3 weeks. Patient data were retrospectively retrieved from the electronic medical records. According to the nadir of the neutrophil count during cisplatin plus S-1 combination therapy, the patients were categorized into grade 3-4 neutropenia and grade 0-2 neutropenia groups. Patient characteristics before the start of chemotherapy between the two study groups were then compared.
Clinical data included sex, age, body weight, body surface area, body mass index, surgery history, liver metastasis, lymph node metastasis, peritoneal metastasis, laboratory data, and cisplatin and S-1 treatment intensity. Furthermore, we compared the laboratory data between the two groups; these laboratory data were speci cally those obtained before cisplatin administration during the treatment course when the neutrophil count reached nadir. We also investigated the correlation between the serum albumin concentration before cisplatin administration as mentioned above and the uctuation of the neutrophil count after cisplatin administration. The uctuation in the neutrophil count was compared using the difference in the logarithm of each value. The optimal cutoff value, which was the limit value of serum albumin concentration anticipated to develop grade 3-4 neutropenia, was calculated by receiver operating characteristic (ROC) curve analysis. The severity of neutrophil count decrease (neutropenia) was graded according to the Common Terminology Criteria for Adverse Events version 5.0 (CTCAE ver. 5.0, Japanese translation JCOG version). A neutrophil count of <1,000/µL indicates grade 3 neutropenia.

Informed consent
This study protocol conformed to the Declaration of Helsinki and was approved by the Ethics Committee of the Jikei University ]. This retrospective observational study was conducted using the opt-out method to post its purpose in the Jikei University School of Medicine. Informed consents were not required for this study because the clinical data of each patient who agreed to treatment by a written consent were kept anonymous during the analysis. For full disclosure, the study details are mentioned in an opt-out document in the Jikei University School of Medicine.

Statistical analysis
Continuous and categorical variables were statistically analyzed by Mann-Whitney's U test and chisquare test, respectively. The clinical data of each patient were compared using the Wilcoxon signed-rank test. Furthermore, the correlation between albumin concentration and neutrophil count uctuation was examined using the Spearman rank correlation coe cient. All statistical data were analyzed using the SPSS ® version 21. The signi cance level of the test was set at 0.05.
Patients in the grade 3-4 neutropenia and grade 0-2 neutropenia groups reached the nadir of neutrophil count after 3.2 ± 2.3 and 3.2 ± 2.6 courses (P = 0.83), and the mean neutrophil count was 696 ± 188/µL and 2283 ± 1074/µL (P < 0.001), respectively. Table 2 shows the laboratory data obtained before cisplatin administration during chemotherapy when the neutrophil count reached nadir. In the grade 3-4 neutropenia and grade 0-2 neutropenia groups, the mean total protein level was 6.13 ± 0.66 and 6.62 ± 0.67 g/dL, and the mean serum albumin concentration was 3.39 ± 0.60 and 3.85 ± 0.59 g/dL, respectively; both values in the grade 3-4 neutropenia group were signi cantly lower than those in the grade 0-2 neutropenia group (total protein: P = 0.009, serum albumin concentration: P = 0.006, Table 2). In patients with grade 3-4 neutropenia from before the treatment to before cisplatin administration in the treatment course during which the neutrophil count reached nadir, the mean total protein level was signi cantly decreased from 6.48 ± 0.50 g/dL to 6.13 ± 0.66 g/dL (P = 0.002; Table 3). Their mean serum albumin concentration also signi cantly decreased from 3.60 ± 0.54 g/dL to 3.39 ± 0.60 g/dL (P = 0.028). Conversely, the total protein and serum albumin concentration levels in the grade 0-2 neutropenia group did not signi cantly decrease. Furthermore, the logarithmic difference of the uctuation of total protein was −0.025 ± 0.035 and −0.003 ± 0.035, while the uctuation of serum albumin concentration was −0.030 ± 0.058 and 0.008 ± 0.035, in grade 3-4 and grade 0-2 neutropenia groups, respectively; the uctuations in the former group were signi cantly less when compared with the latter group (total protein: P = 0.005, serum albumin concentration: P = 0.009; Table 3). Grade 0-2 3.77 ± 0.51 3.85 ± 0.59 0.10 a) 0.008 ± 0.035 ** P < 0.01 *P < 0.05; a) Wilcoxon signed-rank test, b) Mann-Whitney U test Figure 1 illustrates the correlation between the serum albumin concentration before cisplatin administration in the treatment course during which the neutrophil count reached nadir and the neutrophil count uctuation after cisplatin administration. Lower serum albumin concentrations before cisplatin administration were signi cantly correlated with a decrease in neutrophil count after cisplatin administration (r = 0.463, P < 0.001).
According to the ROC curve analysis, a cutoff serum albumin concentration of 3.25 g/dL before cisplatin administration in the treatment course during which the neutrophil count reached nadir was associated with grade 3-4 neutropenia (area under the curve [AUC]: 0.698, sensitivity: 81.3%, speci city: 50.0% ). Out of 24 patients with serum albumin concentration below the cutoff value of 3.25 g/dL before cisplatin administration, 12 developed grade 3-4 neutropenia (odds ratio for grade 3-4 neutropenia: 4.33). In addition, the frequency of grade 3-4 neutropenia below the cutoff value was signi cantly high (P = 0.012; Table 4).

Discussion
Hypoalbuminemia worsens the treatment prognosis of patients with gastric cancer [5,6]; thus, maintaining the nutritional status is necessary to achieve a better treatment outcome. The role of hypoalbuminemia as a risk factor for therapy disruption should be fully understood to help optimize patient management. In the current study, the mean serum albumin concentrations of patients before cisplatin plus S-1 combination therapy were 3.60 ± 0.54 and 3.77 ± 0.51 g/dL in the grade 3-4 neutropenia and grade 0-2 neutropenia groups, of which 71% and 71% had hypoalbuminemia, respectively. Therefore, in this study, the serum albumin concentration of numerous patients with gastric cancer suffering from gastric dysfunction decreased before the start of chemotherapy. Furthermore, considering that cisplatin plus S-1 combination therapy is associated with a high frequency of gastrointestinal adverse events, the patients' nutritional status will worsen after chemotherapy.
Hypoalbuminemia is frequently linked to severe neutropenia [7][8][9][10], but it has not been fully evaluated in serum albumin concentration during anticancer drug administration as the optimal observation point. In this study, by observing the albumin concentration before cisplatin administration in the treatment course during which the neutrophil count reached nadir, we evaluated the albumin concentration effect on neutrophil count uctuation with pharmacokinetic alteration. The total protein and serum albumin concentration levels before cisplatin administration were lower in the grade 3-4 neutropenia group than in the grade 0-2 neutropenia group, and the degree of such reductions from before chemotherapy to before cisplatin administration were signi cantly greater in the grade 3-4 neutropenia group. Total protein includes serum albumin, globulin, and α1 acidic glycoprotein, and serum albumin constitutes the highest proportion; hence, serum albumin is a confounding factor in the reduction of total protein, demonstrating a strong relationship. Therefore, patients manifesting hypoalbuminemia as the treatment progressed had a high incidence of grade 3-4 neutropenia. Furthermore, Lower serum albumin concentrations before cisplatin administration were signi cantly correlated with a decrease in neutrophil count after cisplatin administration. Our ndings suggested that patients with hypoalbuminemia before cisplatin administration during cisplatin plus S-1 combination therapy were highly susceptible to severe neutropenia development.
Thus, the drug property that cisplatin is particularly susceptible to alteration in serum albumin concentration during administration should be investigated. At the end of cisplatin infusion, the proteinbound fraction increases to 90% after 2 h [3,4], indicating the irreversible binding of platinum [11][12][13].
The half-life for the decrease in the concentration values of nonprotein-bound plasma platinum is approximately 23-45 min at the end of the infusion [14,15], with the rapid elimination of free cisplatin from the sera. After approximately 2 h, lterable platinum and cisplatin levels become undetectable [16]. Immediately after the completion of infusion, free cisplatin concurrently affects the pharmacokinetic processes including protein binding, elimination, and tissue distribution. Thus, cisplatin administration to patients with hypoalbuminemia may affect the increase of drug e cacy by the unbound drug fraction increases.
Identifying the border region indicating hypoalbuminemia risk is necessary for patient monitoring. In the current study, the patients with serum albumin concentration below 3.25 g/dL before cisplatin administration, as calculated by ROC curve analysis, showed a signi cantly high incidence of grade 3-4 neutropenia (odds ratio: 4.33); grade 3-4 neutropenia is a frequent and serious adverse effect of cisplatin plus S-1 combination therapy and may progress to febrile neutropenia. Therefore, a serum albumin concentration below 3.25 g/dL before cisplatin administration should be an important parameter for patient monitoring during chemotherapy.
Considering that severe neutropenia is more frequent in Japanese patients with gastric cancer who received cisplatin combination therapies than in foreigners [1,[17][18][19][20], the analysis of grade 3-4 neutropenia and related factors from the perspective of Japanese population is essential. Although hypoalbuminemia as a risk factor for chemotherapy-induced neutropenia requires further investigation regarding its application to non-Japanese patient populations, serum albumin concentration may be one of the safety criteria for proper chemotherapy involving cisplatin.

Conclusions
Patients with gastric cancer undergoing cisplatin plus S-1 combination therapy often developed hypoalbuminemia since before chemotherapy, and those with further decrease of serum albumin concentration after chemotherapy exhibited a signi cantly greater risk of cisplatin-induced neutropenia. Hence, serum albumin concentration needs to be evaluated before each administration of cisplatin to anticipate severe neutropenia development.

Consent for publication
This was a retrospective observational study and carried out by the opt-out method to post a purpose in the Jikei University School of Medicine. Patients were not required to give informed consent for the study because the analysis used anonymous clinical data that were obtained after each patient agreed to treatment by written consent. For full disclosure, the details of the study are mentioned in the opt-out document in the Jikei University School of Medicine.
Availability of data and materials The datasets generated and analyzed during the present study are available from the corresponding author on reasonable request.

Competing interests
Norio Mitsumori who is a coauthor on this study has earned compensation as an outside director of Ootoya Holdings Co. Ltd., and also received study support from Taiho Pharmaceutical Co. Ltd. and Covidien Japan Co. Ltd. This study has no con ict with these companies.

Funding
This research did not receive any speci c grant from funding agencies in the public, commercial, or notfor-pro t sectors.
Authors' contributions NY designed and performed the research and wrote the paper; KA contributed to the analysis; KM contributed to critical revision of the article for important intellectual content; MN provided clinical advice; KT provided the nal approval for this article. Correlation between serum albumin concentration before cisplatin administration and the uctuation of neutrophil count

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. FCSTROBEchecklistv4combined14.docx