Impact of hemoglobin levels on hemoglobin-adjusted carbon monoxide diffusion capacity after chemotherapy for testicular cancer

Abstract

Objectives

We aimed to compare the diffusion capacity of carbon monoxide (DLCO), which was adjusted using the two equations the Cotes method and the Dinakara method, to assess bleomycin-induced lung injury in testicular cancer patients preparing for post-chemotherapy surgery.

Methods

Between November 1990 and October 2018, 89 patients with advanced testicular cancer were recruited into the study. All patients received chemotherapy and underwent DLCO measurements using the single-breath technique prior to surgery for residual tumor removal.

Results

The mean DLCO adjusted for hemoglobin using the Cotes and Dinakara methods was 69.5% and 86.0%, respectively (P < 0.001). According to the Cotes method, adjusted DLCO was severely diminished to below 65% in 40 patients (45%), whereas this proportion was only 16% according to the Dinakara method. We observed a significant correlation between hemoglobin levels and DLCO adjusted using the Cotes method (P < 0.001), but not using the Dinakara method. Four patients received a clinical diagnosis of bleomycin-induced pneumonitis (BIP), and all patients recovered after oral steroid therapy or observation. The DLCO adjusted by either methods was not well correlated with the development of BIP. No patients had major postoperative respiratory complications.

Conclusions

We found that Cotes-adjusted DLCO may be influenced by anemia. We recommend the addition of Dinakara-adjusted DLCO, along with chest computed tomography, for preoperative risk assessment.

Testicular cancer, diffusion capacity of carbon monoxide, bleomycin, DLCO adjusted for hemoglobin

Introduction

Patients with testicular cancer (TC) have a relatively favorable prognosis when treated appropriately with cisplatin-based chemotherapy and surgery (1). Bleomycin is the key drug in standard first-line chemotherapy for TC, which is a combination of bleomycin, etoposide, and cisplatin (BEP therapy), although bleomycin can cause severe and occasionally fatal pulmonary toxicity (1). Bleomycin-induced pneumonitis (BIP) is observed in approximately 10% of patients treated with BEP therapy (2), and the incidence of fatal BIP is reported to be 1–2% (3,4). Following BEP therapy, surgery (i.e., thoracotomy and retroperitoneal lymph node dissection [RPLND]) removes any residual tumor tissue for the management of TC. Several reports suggest that bleomycin-treated patients are at risk of developing postoperative fatal acute respiratory distress syndrome (ARDS) (5,6).

The diffusion capacity of carbon monoxide (DLCO) has been used for monitoring and determining the risk assessment of pulmonary toxicity and complications from bleomycin exposure. However, it remains unclear whether DLCO alone is sufficiently sensitive and specific to operate as a useful prognostic tool (7). Kawai et al. reported that among 10 patients whose DLCO decreased to below 70% of the predicted value after BEP therapy, only two patients developed clinically evident non-fatal BIP (8). In addition, Mckeage et al. showed significantly diminished DLCO in only one of six patients who developed clinically evident BIP, suggesting deficiencies in both the sensitivity and specificity of DLCO as a marker for risk prediction (9).

Hemoglobin (Hb) tightly binds carbon monoxide. DLCO changes can be substantial as a function of Hb concentration. Therefore, the American Thoracic Society (ATS) and the European Respiratory Society (ERS) guidelines recommend Hb adjustment for determining DLCO (10,11).

To calculate DLCO adjusted for Hb, several equations have been proposed (12–14). Among them, the adjustment equation reported by Cotes et al. (12) is commonly used in many countries. In Japan, the ATS/ERS and the Japanese Respiratory Society have recommended this method. In a more recent study, Marrades et al. proposed a new equation for patients with a wide range of anemia (15). Corrected values by this method were similar to the Cotes method. Therefore, the updated ATS/ERS guidelines still recommend adjustment with the Cotes method (11).

Recently, however, several investigators have pointed out the possibility that the Cotes method may result in overestimates of the magnitude of DLCO decrease, especially in anemic patients (16,17). The hematopoietic cell transplantation-specific comorbidity index (HCT-CI) is an established tool for pre-transplant risk assessment of hematologic malignancies (18). The equation for Hb-adjusted DLCO used in the HCT-CI is based on the Dinakara method (19) rather than the Cotes method. Coffey et al. evaluated DLCO in 73 patients with hematologic malignancy and showed that the mean % DLCO adjusted by the Cotes method was significantly lower than that by the Dinakara method (61% and 77%, respectively) (16). Furthermore, this difference was significantly more pronounced in patients with Hb < 12 g/dl (16). These findings are particularly important when considering pre-surgical evaluation of DLCO in chemotherapy-treated TC because this patient population is frequently anemic. There have been no previous studies directly comparing the estimates generated by these two methods.

In this study, we analyzed DLCO adjusted by both the Cotes and Dinakara methods in TC patients who recently completed chemotherapy. Our main aim was to compare DLCO adjusted by these two methods to assess bleomycin-induced lung injury. We also aimed to clarify the impact of Hb levels on the estimates of Hb-adjusted DLCO provided by these two methods.

Patients and methods

Patients and treatment

Between November 1990 and October 2018, 89 patients with advanced TC underwent surgery at Tsukuba University Hospital (TUH) for the removal of residual tumor tissue after chemotherapy, including BEP therapy. Patient characteristics for the study population are shown in Table 1. The median age at diagnosis was 31 years (range, 16‒61 years). The histology of TC was seminoma in seven patients (7.9%) and non-seminoma in 82 patients (92.1%). According to the International Germ Cell Cancer Collaborative Group (IGCCCG) criteria (20), 19 (21.3%), 34 (38.2%), and 36 (40.5%) patients were classified as having good, intermediate, and poor prognosis, respectively. According to the UICC/TNM classification (21), 24 (27.0%) and 65 (73.0%) were classified as stage II and stage III, respectively. Our program for induction chemotherapy was three courses of BEP for good-prognosis patients, and four courses of BEP for intermediate- or poor prognosis patients. To avoid bleomycin-induced pulmonary toxicity, we routinely used EP (etoposide and cisplatin) or VIP (etoposide, ifosfamide, and cisplatin) instead of BEP for patients aged >50 years. Therefore, 88 of 89 patients enrolled in the present study were younger than 50 years. The remaining patient, aged 61 years, was treated with BEP at another hospital, and referred to TUH for second-line chemotherapy and surgery.

Table 1.

Patient characteristics

Age	Years
Range	16–61
Median	31
Type of primary tumor	n (%)
Seminoma	7 (7.9)
Non-seminoma	82 (92.1)
IGCCCG criteria	n (%)
Good	19 (21.3)
Intermediate	34 (38.2)
Poor	36 (40.5)
Clinical stage at presentation (UICC)	n (%)
I	0 (0)
II	24 (27.0)
III	65 (73.0)
Bleomycin dose at initial chemotherapy	n (%)
90 mg	4 (4.5)
180 mg	4 (4.5)
270 mg	57 (64)
360 mg	24 (27)
Second or more chemotherapy	47 (52.8)
Surgery	n (%)
RPLND	77 (87)
Thoracotomy	29 (33)
RPLND+thoracotomy	7 (7.9)
Others	10 (11.2)
Hb level at DLCO measuring	g/dl
Range	6.1–14.6
Median	9.6
	n (%)
<8 g/dL	17 (19.1)
8-10 g/dL	33 (37.1)
>10 g/dL	39 (43.8)

RPLND, retroperitoneal lymph node dissection.

Hb, hemoglobin.

DLCO, diffusion capacity of carbon monoxide.

BEP consisted of 100 mg/m² etoposide and 20 mg/m² cisplatin intravenously on days 1-5 and 30 mg/body bleomycin intramuscularly on days 1, 8 and 15, with recycling on day 22 (22). It is standard practice at our institution to use EP/VIP during the first or second course of treatment for patients with aggressive lung metastases or other pulmonary complications. In some patients, BEP was changed to EP/VIP during induction chemotherapy, when pulmonary toxicity and/or venous thromboembolism was suspected. For VIP therapy, etoposide was reduced to 75 mg/m², and ifosfamide (1.2 g/m² on days 1–5) was used instead of bleomycin. As a result, the total bleomycin dose ranged from 90 mg to 360 mg, as shown in Table 1.

After first-line chemotherapy, 47 patients (52.8%) received second-line chemotherapy or beyond. The most commonly used second-line regimen was TIP, in which 175 to 210 mg/m² of paclitaxel was administrated in place of etoposide, followed by ifosfamide and cisplatin administered at the same dose as for VIP therapy. Thirteen patients (14.6%) needed third-line chemotherapy or beyond. The median number of chemotherapy cycles for all patients was 4.0, while 42 patients (47.2%) received five or more cycles of cisplatin-based chemotherapy.

After completion of chemotherapy, 77 patients (87%) underwent RPLND. One patient needed adjunctive nephrectomy at the time of RPLND and 29 patients (33%) underwent thoracic procedures, including 22 patients (25%) who underwent video-assisted thoracotomy. seven patients (7.9%) underwent open thoracotomy, four patients (4.5%) underwent wedge resection, and three patients (3.4%) underwent mediastinal dissection. A total of seven patients (7.9%) underwent both RPLND and thoracic procedures.

Pulmonary function tests and calculation of Hb-adjusted DLCO

Pulmonary function tests measuring vital capacity (VC) and DLCO were performed after completion of chemotherapy as a preoperative risk assessment measure. DLCO was measured by single-breath diffusion capacity and Hb level was measured within 1 week prior to pulmonary function testing. As shown in Table 1, the median Hb level was 9.6 g/dl (range, 6.1–14.6 g/dl), and 17 patients (19.1%) had severe anemia, with Hb < 8 g/dl. Moderate anemia (Hb levels of 8–10 g/dl) was found in 33 patients (37.1%), and the remaining 39 patients (43.8%) had mild anemia (Hb > 10 g/dl). We used both the Cotes and the Dinakara equations for Hb-adjustment DLCO, as follows (10,19):

Cotes equation:

$$\mathrm{Hb}-\mathrm{adjustedDLCO}=\mathrm{DLCO}\left(\mathrm{10}\mathrm{.22}+\mathrm{Hb}\right)/\left(\mathrm{1}\mathrm{.7Hb}\right)$$

Dinakara equation:

$$\mathrm{Hb}-\mathrm{adjustedDLCO}=\mathrm{DLCO}/\left(\mathrm{0}\mathrm{.06965Hb}\right)$$

Restrictive pulmonary disease was defined as VC less than 80% of predicted. DLCO was expressed as a percent of predicted (%DLCO), according to standard reference equations used by Fred Hutchinson Cancer Research Center.

Statistical analyses

We compared the mean DLCO adjusted by the Cotes and Dinakara methods using a Wilcoxon signed rank test. The Mann-Whitney U test was used for comparison of adjusted DLCO stratified according to levels of Hb. P-values <0.001 were considered statistically significant. The correlation between adjusted %DLCO and Hb levels according to the Cotes and Dinakara methods was calculated using single regression analysis. All statistical analyses were performed using SPSS® 25.0 for Windows® (SPSS Inc., Chicago, IL, USA)

Results

The mean Hb-adjusted %DLCO values for all patients according to the Cotes and Dinakara methods were 69.5% and 86.0%, respectively (P < 0.001). When the Cotes method was used, 73% of patients showed significantly diminished DLCO; 25 patients (28%) had a %DLCO between 65–80%, and 40 patients (45%) had a %DLCO below 65% of the predicted value (Table 2). In contrast, using the Dinakara method, 27% of patients had a %DLCO between 65–80% of predicted, and only 16% had a %DLCO below 65%.

Table 2.

Hemoglobin-adjusted DLCO according to the Cotes and Dinakara methods

%DLCO	Cotes method	Dinakara method
>80%	24 (27%)	51 (57%)
65-80%	25 (28%)	24 (27%)
<65%	40 (45%)	14 (16%)

When the Cotes method was used, 73% of patients showed significantly decreased DLCO, 25 patients (28%) had a %DLCO between 65% and 80%, and an additional 40 patients (45%) showed %DLCO <65% of the predicted value. In contrast, using the Dinakara method, the proportion of patients with DLCO between 65% and 80% and <65% was limited to 27% and 16%, respectively.

The mean %DLCO, as measured by the Cotes method, was significantly lower in anemic patients (Fig. 1). Specifically, the Cotes-adjusted %DLCO values were 76.2% for patients with Hb > 10 g/dl, 65.6% for patients with Hb levels of 8–10 g/dl, and 61.4% for patients with Hb < 8 g/dl. The differences among each group were statistically significant (P < 0.001). On the other hand, the Dinakara method was not affected by Hb level (Fig. 1). The mean %DLCO was 87.1% in patients with Hb > 10 g/dl, 83.6% in patients with Hb levels of 8–10 g/dL, and 88.1% in patients with Hb < 8 g/dl. The Hb-adjusted %DLCO was significantly lower when measured by the Cotes method relative to the Dinakara method in each Hb level group (P < 0.001). In addition, the %DLCO by the Cotes method was significantly correlated with Hb levels (P < 0.001, Fig. 2). The trend was noticed both in patients treated with a total bleomycin dose of 360 mg, and up to 210 mg. In contrast, there was no correlation between Hb levels and %DLCO measured by the Dinakara method.

Figure 1

Mean %DLCO adjusted for Hb using the Cotes and Dinakara methods according to Hb levels.

Patients were divided into three groups according to hemoglobin (Hb) levels at the time when DLCO was measured (Hb >10, 8–10, and <8 g/dL). According to the Cotes methods, there was a significant difference in %DLCO among the three groups. In contrast, no significant differences were observed for %DLCO calculated by the Dinakara method.

Figure 2

Correlation of Hb levels and Hb-adjusted %DLCO according to the Cotes and Dinakara methods.

According to the Cotes method, there was a significant tendency toward decreasing %DLCO with decreasing Hb levels, irrespective of total bleomycin dose. No significant tendency was observed between Hb levels and %DLCO calculated by the Dinakara method.

In addition, we analyzed the potential influence of TIP therapy for %DLCO. Forty-three of 89 patients received TIP therapy. In the Cotes method, mean %DLCO of patients who received TIP therapy and those who did not were 67.7% and 71.4%, respectively. In the Dinakara method, those values were 85.6% and 86.4%, respectively. In both the Cotes and Dinakara methods, there was no statistical difference between patients who received TIP therapy and those who did not (P = 0.247 and 0.883, respectively).

A clinical diagnosis of BIP was made in four patients, three of whom were successfully treated with an oral steroid. In the remaining patient, BIP spontaneously resolved without the need for further treatment. As such, all four patients were able to undergo surgery as planned. Table 3 shows the correlation between %DLCO and various clinical findings, including the development of BIP.

Table 3.

Clinical findings according to %DLCO

	%VC < 80%	Interstitial opacity on chest CT	Clinically evident BIP
Cotes method
>80%	1 (4.2%)	1 (4.2%)	1 (4.2%)
65-80%	3 (12%)	2 (8%)	1 (4.0%)
<65%	13 (32.5%)	7 (17.5%)	2 (5.0%)
Dinakara method
>80%	4 (7.8%)	5 (9.8%)	2 (3.9%)
65-80%	7 (29.2%)	1 (4.2%)	1 (4.2%)
<65%	6 (42.9%)	4 (28.6%)	1 (7.1%)

VC, vital capacity.

CT, computed tomography.

BIP, bleomycin-induced pneumonitis.

Regardless of the adjustment method, the proportion of patients who showed %VC < 80% increased as %DLCO decreased. A similar trend was observed in patients with an interstitial opacity apparent on chest CT. However, the trend was not as clear for the proportion of patients who were clinically diagnosed with BIP. According to the Cotes method, clinically evident BIP was diagnosed in only 5% of patients whose %DLCO was <65%. BIP was also diagnosed in around 5% of patients whose %DLCO levels were 65%-80% or >80%. A similar trend was observed in %DLCO adjusted by the Dinakara method, where BIP was diagnosed in around 5% of patients whose %DLCO levels were 65–80% or >80%.

The proportion of patients with %VC less than 80% increased as %DLCO decreased, regardless of the adjustment method. A similar trend was found for patients with interstitial opacity identified by chest CT. However, the trend was not as clear for the proportion of patients who were clinically diagnosed with BIP. According to the Cotes method, clinically evident BIP was diagnosed in only 5.0% of patients whose %DLCO was below 65%. BIP was also diagnosed in approximately 5% of patients whose %DLCO values were 65–80%, or >80%. A similar trend was observed for %DLCO adjusted by the Dinakara method, where BIP was diagnosed in approximately 5% of patients whose %DLCO were 65–80% or >80%. Pulmonary morbidity after surgery was minimal. No patients required postoperative reintubation. There were no patients who suffered from major post-operative respiratory complications such as severe BIP or ARDS.

Discussion

DLCO has long been used for monitoring and risk assessment of bleomycin-induced lung injury. We evaluated DLCO measured after completion of chemotherapy in 89 TC patients preparing for post-chemotherapy surgery. According to the most widely used equation, %DLCO adjusted for Hb using the Cotes method, 73% of patients showed significantly diminished DLCO, 25 patients (28%) had %DLCO between 65% and 80%, and an additional 40 patients (45%) had %DLCO values below 65%. Although interstitial opacity was revealed by chest CT in 10 patients (11%), a clinical diagnosis of BIP was made in only four patients. After the BIP had resolved, either spontaneously (1 patient) or after administration of oral steroid therapy (3 patients), all patients underwent the planned surgery without major post-operative respiratory complications such as BIP or ARDS.

Several notable findings from this study are worth highlighting. First, as shown in Fig. 1, %DLCO, adjusted by the Cotes method, was significantly diminished in anemic patients. Furthermore, there was a significant correlation between Hb levels and %DLCO adjusted by the Cotes method, as shown in Fig. 2. In contrast, %DLCO adjusted by the Dinakara method was not affected by Hb level (Figs 1 and 2). This might partly explain the observed differences in the proportion of patients with severely diminished %DLCO (below 65% of the predicted value), which was 45% according to the Cotes method compared to only 16% by the Dinakara method. It should be noted that BIP developed in only two out of 40 patients (5%) with severely diminished %DLCO according to the Cotes method. Several previous reports have also pointed out the low specificity of decreased DLCO for predicting the development of BIP (8,9,23). Unfortunately, the method of Hb adjustment was often not described in these previous studies. Nevertheless, the largest published study using the Cotes method showed that none of the 12 patients in their population with a major DLCO decrease (35% or more compared to pretreatment levels) ended up with a clinical diagnosis of BIP (23).

Second, BIP can develop in patients with normal %DLCO (>80% of the predicted value). In the present study, one of 24 patients (4.2%) with %DLCO >80%, according to the Cotes method, developed BIP. Also, two of 51 patients (3.9%) with %DLCO >80% by the Dinakara method developed BIP. Mckeage et al. showed that DLCO was significantly diminished in only one of six patients who developed clinically evident BIP (9). Bell et al. also reported on two cases of BIP without a significant reduction in DLCO (24). When considering the fact that BIP might be fatal, these findings support routine inclusion of chest CT examination in preoperative risk assessment, even for patients with nearly normal DLCO. It is our policy to perform chest CT before surgery, even if no lung metastases were revealed on pretreatment CT.

Third, when comparing the Cotes and Dinakara methods, the latter seemed to be superior as the values were stable and unaffected by Hb levels. However, it should be noted that there were a limited number of studies in TC patients using %DLCO adjusted by the Dinakara method (25,26). Cleary, further investigations are needed to evaluate the routine use of the Dinakara method as a suitable replacement for the Cotes method. As a general rule, omission of bleomycin from the treatment regimen is recommended if the DLCO is significantly decreased during chemotherapy, but the cut-off values for a significant decline vary from >25% to >40% from pretreatment levels (23,27). Recently, Roncolato et al. reported the results of a prospective study of accelerated BEP using a cut-off of >25% reduction (23). They found that treatment decisions for asymptomatic patients that incorporated reduction in DLCO led to 20% of bleomycin doses being omitted, while no patient developed other evidence of BIP. The authors concluded that a > 25% reduction in DLCO appeared too cautious a threshold, and questioned the routine use of the pulmonary function test as a method for monitoring bleomycin toxicity (23).

Fourth, in our study no patients developed major post-operative respiratory complications. This is partly due to the appropriate pre-surgical risk assessment, including a pulmonary function test and chest CT. However, it is noteworthy that all patients except one were younger than 50 years old in this study. For the patient aged ≥50, we used EP/VIP therapy according to IGCCCG criteria. In addition, it is standard protocol at our institution to use of EP/VIP therapy during the first or second course of treatment for patients with aggressive lung metastases or other pulmonary complications. As a result, the bleomycin dose was equal to or less than 270 mg in 73% of our patients. Therefore, patient selection and the lower bleomycin dose might have contributed to the lack of observed postoperative events in the present study. Recently, Calaway et al. also reported that pulmonary morbidity was not increased in 191 IGCCCG good-risk TC patients who underwent RPLND after three courses of BEP (28).

To the best of our knowledge, this is the first report to examine the impact of Hb levels on DLCO values adjusted for Hb using the Cotes and Dinakara methods in TC patients. However, there are two important limitations in the present study. First, the study population is relatively small. Only four patients developed clinically evident BIP. Therefore, we could not draw a conclusion about which DLCO-adjusted method was predictive for the risk of BIP. Second, due to the retrospective design, we could not rule out the possibility of selection bias. However, no prospective study exists in which the association between preoperative DLCO and pulmonary morbidity might be analyzed.

Despite these limitations, our data showed that Cotes-adjusted DLCO may be influenced by anemia. Therefore, we recommend the addition of DLCO adjusted for Hb using the Dinakara method, along with chest CT for preoperative risk assessment, among patients with no other clinical signs or symptoms of BIP.

Conflict of interest statement

None.