Secondary primary cancer in patients with head and neck carcinoma: the differences among hypopharyngeal, laryngeal, and other sites of head and neck cancer.

Secondary primary cancer in patients with head and neck

carcinoma: the differences among hypopharyngeal,

laryngeal, and other sites of head and neck cancer

W-S. LIU,

,

, Department of Radiation Oncology, Kaohsiung Veteran General Hospital, Kaohsiung,

Y-J. CHANG,

, Management Office for Health Data, China Medical University Hospital, Taichung, and

Graduate Institute of Clinical Medicine Science and School of Medicine, College of Medicine, China Medical

University, Taichung, C-L. LIN,

, Management Office for Health Data, China Medical University Hospital,

Taichung, J-A. LIANG,

, Department of Radiation Therapy and Oncology, China Medical University Hospital,

Taichung, and Graduate Institute of Clinical Medicine Science and School of Medicine, College of Medicine, China

Medical University, Taichung, F-C. SUNG,

,

, Management Office for Health Data, China Medical University

Hospital, Taichung, and Department of Public Health, China Medical University, Taichung, I- M. HWANG,

PHD

, Department of Medical Imaging and Radiology, Shu-Zen Junior College of Medicine and Management,

Kaohsiung, & C-H. KAO,

, Graduate Institute of Clinical Medicine Science and School of Medicine, College of

Medicine, China Medical University, Taichung, and Department of Nuclear Medicine and PET Center, China

Medical University Hospital, Taichung, Taiwan

INTRODUCTION

Evidence has shown that patients with head and neck cancer (HNC) have a high risk of developing secondary aerodigestive malignant disease (Deleyiannis & Thomas 1997; Chuang et al. 2008; Lee et al. 2009; Milano et al.

2012). Among these secondary aerodigestive cancers,

oesophageal and lung cancer are the two most frequently documented malignancies (McDonald et al. 1989;

Vaamonde et al. 2003; Chuang et al. 2008; Chen et al.

2010). Chen et al. (2010) reported that patients with secondary oesophageal or lung cancer had a 31–105% excess

risk of death compared with those who did not develop secondary primary cancer (SPC). Rennemo et al. (2008) observed that the prognosis was poor for patients with SPC, and the median survival was only 12 months. Hence, to understand the characteristics of SPC after primary HNC is a vital clinical issue. Numerous efforts have recently been made to identify the incidence of SPC after primary HNC (Chuang et al. 2008; Lee et al. 2009), the risk factors associated with developing SPC (Milano et al. 2012), and the risks among different locations of primary tumours (Chen et al. 2010; Milano et al. 2012).

However, data documenting the lag of development among different SPCs are insufficient (Aydiner et al.

2000). Numerous studies have demonstrated the correlation between metabolic disease (e.g. diabetes, hypertension, and hyperlipidaemia) and malignant diseases (ARB Trialists Collaboration 2011; Bhaskaran et al. 2012; Lee

et al. 2012). According to our research, no study has discussed this issue regarding HNC. In addition to organ

preservation (Lefebvre et al. 2012) and evidence of sensitivity to chemotherapy and radiotherapy (Pointreau

et al. 2009; Liu et al. 2010), the treatment policy of hypopharyngeal and laryngeal cancer was different to that of other HNC sites, such as the oral cavity. Milano et al.

(2012) found that the index tumours of the hypopharynx

and supraglottic larynx had a significantly high incidence

(P < 0.0001) of SPC in the lungs. However, we did not

know whether this finding could be observed in Asia.

Therefore, in this study, we aimed to identify the differing characteristics of SPC among hypopharyngeal, laryngeal, and other HNC sites.

METHODS

Ethical considerations

We confirm that all identification information of the study patients was scrambled, and that the data were analysed anonymously. This study was approved by the Ethics Review Board at China Medical University (CMU-REC-101-012).

Data sources

The sampling cohort was obtained from the National Health Insurance Research Database (NHIRD) of Taiwan.

The National Health Insurance (NHI) programme is a government-run, single-payer entity administered by the Bureau of National Health Insurance (BNHI). In March 1995, the Taiwan Department of Health integrated 13

health insurance schemes into a universal insurance programme, and approximately 99% of the 23.74 million

residents in Taiwan were enrolled in the NHIRD (Cheng 2009). This study uses the NHIRD subset data files of catastrophic illnesses, and the entire medical claims of insurants. We were able to link NHI data to other data sets by using scrambled patient identification to obtain the chronological medical history of patients. The International Classification of Diseases, Ninth Revision of

Clinical Modification (ICD-9-CM) was used to identify diagnoses of diseases in the claims data. Data files are linked with scrambled identification to protect patient privacy.

Study patients

Patients with newly diagnosed hypopharyngeal and laryngeal cancer (ICD-9-CM codes 148 and 161, respectively)

from 1997 to 2010, identified from the registry of the

Catastrophic Illnesses Patient Database (CIPD), were considered the study cohort. The date for cancer registration

was defined as the index date. Patients with a history of

cancer (ICD-9-CM codes 140–208) before the index date,

or aged less than 20 years at the index date were excluded.

A comparison cohort comprising patients with newly diagnosed HNC (excluding hypopharyngeal and laryngeal) (ICD-9-CM 140–147, 149) from 1997 to 2010 was

also identified from the CIPD. The comparison cohort was randomly selected by frequency matching at a 1:1 ratio for age, sex, and index year in the study cohort.

Outcome measures

Subsequent malignancies (secondary cancer) (ICD-9-CM 150–194, 200–208) after the index date were identified as outcomes of interest. Metastatic cancer, defined by relapses within 6 months, was not included in the analysis.

The person-years of follow-up were calculated from

the index date to the subsequent cancer diagnoses, withdrawal from insurance because of death, loss to follow-up,

or 31 December 2010, whichever came first. The diseases before the index date that were considered comorbidities included diabetes (ICD-9-CM 250), hypertension (ICD- 9-CM 401–405), and hyperlipidaemia (ICD-9-CM 272).

Statistical analysis

Group differences between the study and comparison

cohorts were assessed using the chi-squared test for categorical variables, and the Wilcoxon two-sample test for

continuous variables. Incidence rates of secondary cancer were estimated in both cohorts stratified by demographic and comorbidity statuses. The incidence rate ratio (IRR) and 95% confidence interval (CI) were estimated using the Poisson regression model. Cox’s proportional-hazards regression model was used to assess the adjusted risk of the incidence of secondary cancer between the study and comparison cohorts. Incidence rates and the hazard ratio (HR) for site-specific secondary cancer were further estimated.

All analyses were performed using the SAS statistical package (version 9.1; SAS Institute, Cary, NC, USA).

A P value of 0.05 was considered significant.

RESULTS

This study included 5914 patients with incident

hypopharyngeal and laryngeal cancer in the study cohort,

and another 5914 patients with incident HNC excluding hypopharyngeal and laryngeal cancer in the comparison

cohort from 1997 to 2010 (Table 1). Demographic characteristics showed that there was no difference between the

two cohorts in age and sex. Both study and comparison cohorts were predominantly male (96.0%) and aged 40–64 years (approximately 59%). The mean follow-up period of the study cohort was significantly shorter than that of the comparison cohort (2.46 vs. 2.71 years, P < 0.0001). The baseline comorbidity showed that the study cohort had higher proportions of pre-existing hypertension (19.1% vs.

16.9%, P = 0.002) and hyperlipidaemia (5.43% vs. 4.58%, P = 0.035). The overall IRR of secondary cancer was 68%

higher in the study cohort than in the comparison cohort (23.9 vs. 14.2 per 1000 person-years, IRR = 1.68, 95% CI = 1.52–1.85) with an adjusted HR of 1.62 (95% CI = 1.37–

1.92) (Table 2). Sex-specific analysis showed that the study cohort to comparison cohort rate ratio was higher for men than for women, and showed a 65% increase of secondary cancer risk in men compared with women. The IRR was highest in the younger subgroup aged under 40 years (IRR

= 2.63, 95% CI = 1.56–4.45), and was lowest in the older subgroup aged 65 years or over (IRR = 1.29, 95% CI = 1.10–1.52). However, the adjusted HR was 1.74 (95% CI = 1.05–2.90) for older patients compared with patients ≤39 years of age. The incidence of secondary cancer measured by comorbidity status showed that patients with preexisting diabetes and hypertension in the study cohort

exhibited a significantly higher incidence. Furthermore, patients with pre-existing hypertension also showed a significantly elevated risk (adjusted HR = 1.32, 95% CI = 1.04–1.67). The analysis results of type-specific secondary cancer are shown in Table 3. The study cohort had a higher IRR than the comparison cohort for oesophageal cancer (IRR = 3.22, 95% CI = 2.40–5.03), hepatoma (IRR = 1.65, 95% CI = 1.46–1.87), lung cancer (IRR = 1.74, 95% CI

= 1.78–2.26), and prostate cancer (IRR = 1.18, 95% CI =

1.04–1.34). However, the study cohort had a lower IRR

than the comparison cohort in thyroid cancer (IRR = 0.66, 95% CI = 0.57–0.77). Compared with the comparison cohort, the adjusted HR of developing oesophageal cancer and lung cancer was 3.47-fold and 1.89-fold, respectively, which was statistically significantly higher for patients with hypopharyngeal and laryngeal cancer patients (respective 95% CIs = 2.40–5.03 and 1.28–2.81).

DISCUSSION

Evidence has shown that survivors of head and neck malignant diseases have a high incidence of secondary primary aerodigestive tract cancers (Leon et al. 1999;

Chuang et al. 2008; Chen et al. 2010; Milano et al. 2012).

Numerous researchers have attempted to identify the type of index tumours in head and neck malignancies that might have a higher incidence of SPC. Therefore, physicians could design individual follow-up examination protocols according to different head and neck tumours.

Milano et al. (2012) analysed 61 883 patients with primary HNC from the Surveillance, Epidemiology, and End

Results database. They found that the index tumours of the hypopharynx (22.8%) and supraglottic larynx (27.1%) had a significantly high incidence (P < 0.0001) of SPC in the lungs. Chen et al. (2010) calculated the standardised

incidence ratios (SIRs) of secondary primary oesophageal and lung cancer from the Taiwan Cancer Registry for the

index tumours of primary HNC. They observed that the incidence of oesophageal cancer had increased in oral/

pharyngeal and laryngeal cancer patients (SIRs 8.71 and 4.65, respectively). The incidence of lung cancer also increased in laryngeal and oral/pharyngeal cancer patients (SIRs 2.05 and 1.56, respectively). In this study, the index tumours of hypopharyngeal and laryngeal cancers showed a significantly higher incidence of oesophageal (IRR 3.22, 111 vs. 38 events) and lung (IRR 2.01, 71 vs.

39 events) cancers compared with other sites of head

and neck tumours (Table 3). These two index tumours

(hypopharyngeal and laryngeal cancers) were the same

as or similar to Western studies, such as Milano et al.

(2012) and Deleyiannis and Thomas (1997). However, although in the same area (Taiwan), Chen et al. (2010) did not obtain similar results. They found that the oral/pharyngeal index tumours had the highest SIR for oesophageal cancer because of two possible reasons. First, the duration of the data sources was different. The registry period for Chen’s study was from 1997 to 2003, and for our study it was from 1997 to 2010. We knew that the order of cancer incidence can change according to different registration periods in Taiwan (Taiwan Cancer Registry 2012). Second, the selection method was different. Chen included oral and pharyngeal (contained hypopharyngeal) cancers in the same category, whereas we divided hypopharyngeal and laryngeal cancer from other HNCs.

Our study shows that the most frequent SPC after the index tumour of HNC (ICD-9-CM codes 140–149, 161) is oesophageal cancer (Table 3). This differs from the results of Western studies, where the most frequent SPC was lung cancer (Deleyiannis & Thomas 1997; Vaamonde et al. 2003; Chuang et al. 2008; Milano et al. 2012). For example, Chuang et al. (2008) published a study composed of 99 257 patients from 13 cancer registries. The majority of cases were from Europe (72.6%). They found that the most common SPC was lung cancer, which contributed a cumulative risk of 13% in 20 years. For the Asia study, Chen et al. (2010) reported that oesophageal cancer was the most frequent SPC after HNC in Taiwan. The differences in carcinogenesis that induced head and neck

tumours between Western and Asian cultures (Taiwan Cancer Registry 2012) may be a factor that contributes to these different results (Ko et al. 1995; Chen et al. 2010).

Various authors have used different criteria to categorise the time intervals between the index date and SPC.

For example, Chen et al. (2010) used less than 1 year and 1–5 years. Chan et al. (2011) conducted a comparative study to identify the role of

F-FDG PET/CT and

3.0-T whole-body magnetic resonance images to detect

distant metastases and SPC in patients with untreated oropharyngeal and hypopharyngeal

carcinoma. They

observed 10 (9.7%) synchronous SPCs in 103 patients.

Among them, the most frequent type of synchronous malignant disease was oesophageal cancer (5 in 10).

According to our research, no study has discussed the pattern of synchronous and metachronous SPCs in HNC.

Our results show that the study cohort had a significantly higher incidence of SPC than did the comparison cohort, as evaluated by the IRR method. These tumours included the oesophagus (IRR = 3.22), lung (IRR = 2.01), and hepatoma (IRR = 1.65) (Table 3). Hence, in addition to that other authors reported SPC in the oesophagus and lungs (McDonald et al. 1989; Deleyiannis & Thomas 1997; Vaamonde et al. 2003; Chuang et al. 2008; Chen et al. 2010), we should examine the possibility of secondary hepatoma when staging work-ups and follow-ups

for hypopharyngeal and laryngeal cancer patients. This finding may not be duplicated in the West, but we should be aware of this possibility because hepatoma is endemic in Taiwan. The incidence of secondary hepatoma (adjusted HR = 1.59, 95% CI = 0.98–2.58) did not reach a significant difference between the two cohorts when evaluated using the HR (Table 3). Secondary oesophageal cancer was the only cancer in which the study cohort had a significantly higher incidence than the comparison cohort (adjusted HR = 2.4). When evaluating metachronous tumours between the two cohorts, we observed that

oesophageal (adjusted HR = 3.99) and lung (adjusted HR = 1.86) cancers had a higher incidence in the study cohort.

From the NHIRD, we could not obtain information

regarding personal habits (smoking, alcohol use, and betel nut chewing), viral risk factors (e.g. Human papillomavirus and Epstein–Barr virus), and treatment methods.

Hence, the limitation of this study is that we could not exclude these confounding factors.

In our study, the hypopharyngeal and laryngeal cancer patients had a significantly higher ratio of hypertension

than other HNC site patients (adjust HR = 1.32; Table 2). The other two metabolic diseases

(diabetes and hyperlipidaemia)

did not show a significant difference between the two cohorts. Numerous studies have recently discussed the correlation between metabolic disease (diabetes, hypertension, and hyperlipidaemia) and malignant diseases (ARB Trialists Collaboration 2011; Chiou et al.

2011; Pelucchi et al. 2011; Rosato et al. 2011; Bhaskaran et al. 2012; Hayashi et al. 2012; Lee et al. 2012; Liao et al.

2012). Chiou et al. (2011) evaluated a large population, and found that diabetes patients had a higher incidence of cancer (pancreatic, liver, uterine, urinary tract, and lung) compared with non-diabetic populations (18.1%

vs. 16.3%). Pelucchi et al. (2011) and Hayashi et al.

(2012) have discovered that hyperlipidaemia patients

had a greater risk of prostate cancer than did the nonhyperlipidaemia population. For hypertension, certain

research groups have attempted to determine whether the use of angiotensin receptor blockers influences cancer incidence (ARB Trialists Collaboration 2011; Bhaskaran et al. 2012). However, according to our research, no study has focused on the correlation between metabolic diseases and HNC. We observed that the study cohort had a significantly higher ratio of hypertensive diseases compared

with the comparison cohort (Table 2). We cannot explain further based on this database or previous research. This requires further investigation and study.

CONCLUSION

The incidence of SPC in the study cohort (hypopharyngeal

and laryngeal cancer) was significantly higher than in the

comparison cohort (other HNCs). The most frequent SPCs

in both cohorts were oesophageal, lung, and hepatocellular

carcinoma. The study cohort had a significantly higher

incidence of these three SPCs compared with the comparison

cohort. When evaluated using adjusted HRs, the study

cohort had a significantly higher incidence of oesophageal

and lung cancer. For synchronous SPC, the study cohort

had a significantly higher adjusted HR in oesophageal

cancer. For metachronous SPC, the study cohort had a

significantly higher adjusted HR in both oesophageal and lung cancer. Based on our results, whether an additional examination of oesophageal fibroscopy or oesophagogram for staging follow-up examinations could benefit hypopharyngeal and laryngeal cancer patients in Taiwan

should be examined. Whether an additional examination

of high-resolution computed tomography chest scans for

follow-up examinations could benefit these patients may

require further investigation.